instance_id
stringlengths
21
53
repo
stringclasses
188 values
language
stringclasses
1 value
pull_number
int64
20
148k
title
stringlengths
6
144
body
stringlengths
0
83.4k
created_at
stringdate
2015-09-25 03:17:17
2025-07-10 16:50:35
problem_statement
stringlengths
188
240k
hints_text
stringlengths
0
145k
resolved_issues
listlengths
1
6
base_commit
stringlengths
40
40
commit_to_review
dict
reference_review_comments
listlengths
1
62
merged_commit
stringlengths
40
40
merged_patch
stringlengths
297
9.87M
metadata
dict
skypilot-org__skypilot-5191@1186d15
skypilot-org/skypilot
Python
5,191
[k8s] Check k8s dependencies on `sky check`
Fixes #4796 <!-- Describe the changes in this PR --> Check whether `socat` and `netcat` are properly downloaded when trying to use Kubernetes as a cloud. ![sky_check](https://github.com/user-attachments/assets/855a8f9b-c3ac-4e06-a7b6-8095d6c31fea) Also handles the case when relevant dependencies are installed an...
2025-04-11T10:36:40Z
[k8s] Raise socat/netcat dependency requirements in `sky check` Currently we ask users to install netcat/socat when they run `sky launch`. Worse, if neither of the dependencies is installed, our error messages do not tell the user that both need to be installed, making it a sequential process: ``` sky check kubernetes ...
+1, users ran into this. I also ran into this recently, this would be helpful to fix. could I try to tackle this? Please feel free to take over this #4807. @kyuds : )
[ { "body": "Currently we ask users to install netcat/socat when they run `sky launch`. Worse, if neither of the dependencies is installed, our error messages do not tell the user that both need to be installed, making it a sequential process:\n```\nsky check kubernetes\n# Passes\nsky launch -c test --cloud k8s -...
b4202948016cdf48a5939ea9bf1769a2d31f73bf
{ "head_commit": "1186d157a56c6d2eb016d4756489d552d4443e18", "head_commit_message": "implementation complete, need to check for active k8s with post-removed socat/netcat", "patch_to_review": "diff --git a/sky/clouds/kubernetes.py b/sky/clouds/kubernetes.py\nindex 268c71f0cdf..6a2996da2e3 100644\n--- a/sky/clouds/...
[ { "diff_hunk": "@@ -2148,32 +2148,33 @@ def fill_ssh_jump_template(ssh_key_secret: str, ssh_jump_image: str,\n return content\n \n \n-def check_port_forward_mode_dependencies() -> None:\n- \"\"\"Checks if 'socat' and 'nc' are installed\"\"\"\n+def check_port_forward_mode_dependencies(\n+ raise_err...
8da21d04e5215d41ae01d4bbc629b4b2240455ec
diff --git a/sky/clouds/kubernetes.py b/sky/clouds/kubernetes.py index 268c71f0cdf..bd7ceab96bc 100644 --- a/sky/clouds/kubernetes.py +++ b/sky/clouds/kubernetes.py @@ -658,6 +658,14 @@ def _make(instance_list): def _check_compute_credentials(cls) -> Tuple[bool, Optional[str]]: """Checks if the user has a...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Dependency Updates & Env Compatibility" }
skypilot-org__skypilot-5098@76905ee
skypilot-org/skypilot
Python
5,098
[UX] Skip `STOPPED` cluster when calling `sky stop`
<!-- Describe the changes in this PR --> Fixes #4090 Simple fix to skip clusters that have `ClusterStatus.STOPPED` <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the relevant ones): - [X] Code formatting: ...
2025-04-02T09:03:41Z
[UX] Skip `STOPPED` cluster when calling `sky stop` <!-- Describe the bug report / feature request here --> We should filter out those clusters in `STOPPED` status when calling `sky stop`. e.g. for the example below, `sky stop -a` should only stop 2 clusters `lmf-l4-4-machine-image` and `lmf-gcp-l4-4`. ```bash ...
[ { "body": "<!-- Describe the bug report / feature request here -->\r\n\r\nWe should filter out those clusters in `STOPPED` status when calling `sky stop`. e.g. for the example below, `sky stop -a` should only stop 2 clusters `lmf-l4-4-machine-image` and `lmf-gcp-l4-4`.\r\n\r\n```bash\r\n$ sky status\r\nCluster...
515e5df7c14cfd425f606de20d4b96d6a897ee7d
{ "head_commit": "76905ee3a93f6dd236a157468f2384e627c7c0da", "head_commit_message": "ignore condition when down", "patch_to_review": "diff --git a/sky/cli.py b/sky/cli.py\nindex 9a8ce8c7267..7d680dd5c45 100644\n--- a/sky/cli.py\n+++ b/sky/cli.py\n@@ -3065,6 +3065,7 @@ def _down_or_stop_clusters(\n rec...
[ { "diff_hunk": "@@ -3065,6 +3065,7 @@ def _down_or_stop_clusters(\n record['name']\n for record in all_clusters\n if controller_utils.Controllers.from_name(record['name']) is None\n+ and (down or record['status'] != status_lib.ClusterStatus.STOPPED)", "line": n...
f04353d83b257462c9458bfd550b1abe0e86a5c1
diff --git a/sky/cli.py b/sky/cli.py index 9a8ce8c7267..d4cd327f902 100644 --- a/sky/cli.py +++ b/sky/cli.py @@ -3061,10 +3061,15 @@ def _down_or_stop_clusters( 'Letting --all take effect.') # We should not remove controllers when --all is specified. # Otherwise, it would be very easy...
{ "difficulty": "low", "estimated_review_effort": 2, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-5096@be11ac1
skypilot-org/skypilot
Python
5,096
[API server] Set default min resources and warning on low resources
<!-- Describe the changes in this PR --> close https://github.com/skypilot-org/skypilot/issues/5093 <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the relevant ones): - [x] Code formatting: install pre-commi...
2025-04-02T06:45:55Z
[API server] Set default resources requests in helm deployment One of our user use kata-container as container runtime, which set a default tiny VM size if the Pod has no resources specified. We should set a default resources request in our chart.
[ { "body": "One of our user use kata-container as container runtime, which set a default tiny VM size if the Pod has no resources specified. We should set a default resources request in our chart.", "number": 5093, "title": "[API server] Set default resources requests in helm deployment" } ]
8f1e5066538bdbdfe73caaaa4330a4ef3304daab
{ "head_commit": "be11ac1d03bc7ec0ef7677d90ccaaed7e577bfdc", "head_commit_message": "Update charts/skypilot/templates/_helpers.tpl\n\nCo-authored-by: Zhanghao Wu <zhanghao.wu@outlook.com>", "patch_to_review": "diff --git a/charts/skypilot/templates/NOTES.txt b/charts/skypilot/templates/NOTES.txt\nnew file mode 10...
[ { "diff_hunk": "@@ -0,0 +1,42 @@\n+.. _sky-api-server-troubleshooting:\n+\n+Troubleshooting SkyPilot API Server\n+===================================\n+\n+If you're unable to run SkyPilot API server, this guide will help you debug common issues.\n+\n+If this guide does not help resolve your issue, please reach ...
31f4eb645030536b631b82683fe111f8cacd9569
diff --git a/charts/skypilot/templates/NOTES.txt b/charts/skypilot/templates/NOTES.txt new file mode 100644 index 00000000000..85105a04239 --- /dev/null +++ b/charts/skypilot/templates/NOTES.txt @@ -0,0 +1,3 @@ +{{- if not .Values.apiService.skipResourceCheck }} +{{- include "skypilot.checkResources" . }} +{{- end }} d...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Performance Optimizations" }
skypilot-org__skypilot-5091@ada5dad
skypilot-org/skypilot
Python
5,091
[fast] don't skip setup if we do provisioning
Fixes #5089. <!-- Describe the changes in this PR --> <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the relevant ones): - [x] Code formatting: install pre-commit (auto-check on commit) or `bash format.s...
2025-04-02T02:42:42Z
[Jobs] Controller does not get a newly setup cloud due to missing of re-setup of cloud dependencies We have `fast=True` for controller, which does not run setup again on the controller when the controller is UP, causing the newly configured clouds are not enabled on the controller. This is a big issue for controllers ...
[ { "body": "We have `fast=True` for controller, which does not run setup again on the controller when the controller is UP, causing the newly configured clouds are not enabled on the controller.\n\nThis is a big issue for controllers on Kubernetes clusters, as the controller never autostop, i.e. it will never be...
b7d83e1c79e8e8c174262840de92dcd0c879fc42
{ "head_commit": "ada5dad79ab181f1cefe0df6f3bed2aad8794b68", "head_commit_message": "fix other backend.provision call", "patch_to_review": "diff --git a/sky/backends/backend.py b/sky/backends/backend.py\nindex 210496cd54c..e291944e48c 100644\n--- a/sky/backends/backend.py\n+++ b/sky/backends/backend.py\n@@ -1,6 +...
[ { "diff_hunk": "@@ -66,15 +66,17 @@ def provision(\n cluster_name: Name of the cluster to provision. If None, a name will\n be auto-generated. If the name refers to an existing cluster,\n the existing cluster will be reused and re-provisioned.\n- retry_unti...
307551300baaba04a5fd08c58aa5fc5d1d852352
diff --git a/sky/backends/backend.py b/sky/backends/backend.py index 210496cd54c..ca40cb763da 100644 --- a/sky/backends/backend.py +++ b/sky/backends/backend.py @@ -1,6 +1,6 @@ """Sky backend interface.""" import typing -from typing import Dict, Generic, Optional +from typing import Dict, Generic, Optional, Tuple ...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-5066@34e607a
skypilot-org/skypilot
Python
5,066
[DO] Refactor DigitalOcean credential path handling
Introduced `get_credentials_path` to encapsulate credential path logic, improving code readability and maintainability. Replaced direct access to `CREDENTIALS_PATH` with the new method in relevant parts of the code, ensuring consistency in handling credentials. Fix https://github.com/skypilot-org/skypilot/issues/506...
2025-03-28T18:41:04Z
sky jobs launched failed One of our users report jobs launch failure after upgrade to latest nightly: ``` E 03-28 10:45:10 sdk.py:1468] === Traceback on SkyPilot API Server === E 03-28 10:45:10 sdk.py:1468] Traceback (most recent call last): E 03-28 10:45:10 sdk.py:1468] File "/opt/conda/lib/python3.10/site-packages...
I think we should merge #5061 immediately to resolve the issue on master, and then we can follow up with #5066.
[ { "body": "One of our users report jobs launch failure after upgrade to latest nightly:\n\n```\nE 03-28 10:45:10 sdk.py:1468] === Traceback on SkyPilot API Server ===\nE 03-28 10:45:10 sdk.py:1468] Traceback (most recent call last):\nE 03-28 10:45:10 sdk.py:1468] File \"/opt/conda/lib/python3.10/site-packages...
9e6cef22ae679a1fef4539cbfc02a6d5cdc405ed
{ "head_commit": "34e607ad0bf7723b79b1493e6fc8f7362e5e9e40", "head_commit_message": "Merge branch 'master' into fixdocreds", "patch_to_review": "diff --git a/sky/clouds/do.py b/sky/clouds/do.py\nindex 41b7bc49bc5..c4d74def67f 100644\n--- a/sky/clouds/do.py\n+++ b/sky/clouds/do.py\n@@ -280,12 +280,14 @@ def _check...
[ { "diff_hunk": "@@ -280,12 +280,14 @@ def _check_compute_credentials(cls) -> Tuple[bool, Optional[str]]:\n return True, None\n \n def get_credential_file_mounts(self) -> Dict[str, str]:\n- if do_utils.CREDENTIALS_PATH is None:\n+ if do_utils.get_credentials_path() is None:", "line"...
793945ea1ba2a3eb54885057683df48758c144cc
diff --git a/sky/clouds/do.py b/sky/clouds/do.py index 41b7bc49bc5..e83f0bd470d 100644 --- a/sky/clouds/do.py +++ b/sky/clouds/do.py @@ -280,13 +280,12 @@ def _check_compute_credentials(cls) -> Tuple[bool, Optional[str]]: return True, None def get_credential_file_mounts(self) -> Dict[str, str]: - ...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-5030@0acc040
skypilot-org/skypilot
Python
5,030
[Core] Support setting sudo password when deploying remote clusters
<!-- Describe the changes in this PR --> Fix #4795 - Update doc to recommend using passwordless sudo when deploying remote clusters with `sky local up` - If passwordless sudo is not available, support setting sudo password when deploying remote clusters with `sky local up` <!-- Describe the tests ran --> <!...
2025-03-25T13:12:36Z
[k8s] Allow sudo with password for `sky local up` Currently, if a user's sudo access requires password on the remote cluster, `sky local up` will fail. We can either: 1. Add support for password with sudo (better) 2. Add docs for setting up passwordless sudo on remote instance
One way to support it is to just prompt the user to enter the password Discussion summary: 1. We prefer the user to set up passwordless sudo in the remote server 2. If 1 cannot be done, we request the user to set a password in `sky local up` options
[ { "body": "Currently, if a user's sudo access requires password on the remote cluster, `sky local up` will fail. We can either:\n1. Add support for password with sudo (better)\n2. Add docs for setting up passwordless sudo on remote instance", "number": 4795, "title": "[k8s] Allow sudo with password for ...
052e0ea3b129f208a82fd9d6bddbfd39d389e831
{ "head_commit": "0acc04016ce74a5cc194f8afa70237464160e47b", "head_commit_message": "update doc", "patch_to_review": "diff --git a/docs/source/reservations/existing-machines.rst b/docs/source/reservations/existing-machines.rst\nindex a54af88f4c4..0a0f4079ae0 100644\n--- a/docs/source/reservations/existing-machine...
[ { "diff_hunk": "@@ -101,9 +117,19 @@ cleanup_agent_node() {\n local NODE_IP=$1\n echo -e \"${YELLOW}Cleaning up node $NODE_IP...${NC}\"\n run_remote \"$NODE_IP\" \"\n+ # Create temporary askpass script\n+ ASKPASS_SCRIPT=\\$(mktemp)\n+ trap 'rm -f \\$ASKPASS_SCRIPT' EXIT INT TERM...
e27d2867b079ef21b6bf507e3b7c3760bd9656b9
diff --git a/docs/source/reservations/existing-machines.rst b/docs/source/reservations/existing-machines.rst index a54af88f4c4..882e02510be 100644 --- a/docs/source/reservations/existing-machines.rst +++ b/docs/source/reservations/existing-machines.rst @@ -47,7 +47,8 @@ Prerequisites **Remote machines (your cluster, o...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "New Feature Additions" }
skypilot-org__skypilot-5015@eb3023c
skypilot-org/skypilot
Python
5,015
[core] optimizer/provisioner fails over for cloud with expired credentials
<!-- Describe the changes in this PR --> Fix #4373 - For Kubernetes with expired credentials, auto-exclude the contexts - For invalid credentials of aws, azure, and gcp after enabled in cache, failover to the other clouds for provisioning <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are par...
2025-03-23T10:39:39Z
[Core] Expired credentials causes unexpected failure of `sky launch` <!-- Describe the bug report / feature request here --> Many cloud credentials have expiration, e.g. AWS SSO, or kubeconfig with gcloud authentication. Expiration of these credentials means, although the cloud is enabled in the cache, `sky launch` ...
Related [issue](https://github.com/skypilot-org/skypilot/issues/4360) Root cause: The enabled clouds list will be cached after `sky check`, while for the following operations, they only refresh the list if no clouds are enabled in the cache or refresh the ones not enabled in the cache. Solution: To fix this issue, we ...
[ { "body": "<!-- Describe the bug report / feature request here -->\r\n\r\nMany cloud credentials have expiration, e.g. AWS SSO, or kubeconfig with gcloud authentication. Expiration of these credentials means, although the cloud is enabled in the cache, `sky launch` will go to those cloud and try to launch the c...
633e16611f2f858dc27c9eae2f410811e0bc714c
{ "head_commit": "eb3023ccc123bf57d2cc24aceacd543623b99abe", "head_commit_message": "address comments", "patch_to_review": "diff --git a/sky/adaptors/gcp.py b/sky/adaptors/gcp.py\nindex a12a41d07c7..6ed2ee5e56a 100644\n--- a/sky/adaptors/gcp.py\n+++ b/sky/adaptors/gcp.py\n@@ -68,6 +68,13 @@ def credential_error_e...
[ { "diff_hunk": "@@ -1089,6 +1089,22 @@ def _lambda_handler(blocked_resources: Set['resources_lib.Resources'],\n FailoverCloudErrorHandlerV2._default_handler(\n blocked_resources, launchable_resources, region, zones, error)\n \n+ @staticmethod\n+ def _aws_handler(blocked_resourc...
5efdf79971a86c1f2ce2e05270402b58763fe178
diff --git a/sky/adaptors/gcp.py b/sky/adaptors/gcp.py index a12a41d07c7..6ed2ee5e56a 100644 --- a/sky/adaptors/gcp.py +++ b/sky/adaptors/gcp.py @@ -68,6 +68,13 @@ def credential_error_exception(): return exceptions.DefaultCredentialsError +@common.load_lazy_modules(_LAZY_MODULES) +def auth_error_exception(): ...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-4908@f5eaa49
skypilot-org/skypilot
Python
4,908
[Core] Handle mid-sequence chunking in log streaming
Closes #4905. Our chunking may chunk in the middle of a UTF-8 sequence, which would cause decoding to break. We now store the undecoded bytes in a buffer and wait to receive the next chunk before trying to decode again.
2025-03-06T19:57:20Z
decode_rich_status can chunk in the middle of utf-8 sequences On some occasions my logs are not retrievable and they end with: ``` File "/home/user/.keybotic/venvs/skypilot/bin/sky", line 8, in <module> sys.exit(cli()) File "/home/user/.keybotic/venvs/skypilot/lib/python3.8/site-packages/click/core.py", line 1161...
[ { "body": "On some occasions my logs are not retrievable and they end with:\n```\nFile \"/home/user/.keybotic/venvs/skypilot/bin/sky\", line 8, in <module>\n sys.exit(cli())\n File \"/home/user/.keybotic/venvs/skypilot/lib/python3.8/site-packages/click/core.py\", line 1161, in __call__\n return self.main...
86e242f811bbab00fc896a3f6a6b3dd591eb6626
{ "head_commit": "f5eaa49b43d07f16d42892132addc0b5a89cb66f", "head_commit_message": "Handle mid-sequence chunking", "patch_to_review": "diff --git a/sky/utils/rich_utils.py b/sky/utils/rich_utils.py\nindex 54e9672f0ac..2a82d5ce679 100644\n--- a/sky/utils/rich_utils.py\n+++ b/sky/utils/rich_utils.py\n@@ -230,15 +2...
[ { "diff_hunk": "@@ -0,0 +1,85 @@\n+import unittest\n+from unittest import mock\n+\n+import requests\n+\n+from sky.utils import rich_utils\n+\n+\n+class RichUtilsTest(unittest.TestCase):\n+\n+ def test_decode_rich_status_with_split_utf8(self):\n+ \"\"\"Test that decode_rich_status handles UTF-8 charact...
bc33bced9ba01b14ea12bcb5d7a0216a3bf3a5aa
diff --git a/sky/utils/rich_utils.py b/sky/utils/rich_utils.py index 54e9672f0ac..2f4dae30083 100644 --- a/sky/utils/rich_utils.py +++ b/sky/utils/rich_utils.py @@ -230,15 +230,53 @@ def decode_rich_status( decoding_status = None try: last_line = '' + # Buffer to store incomplete UTF-8 bytes b...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-4846@935d3c8
skypilot-org/skypilot
Python
4,846
[Core] Exit with non-zero code on launch/exec/logs/jobs launch/jobs logs
Closes https://github.com/skypilot-org/skypilot/issues/4599. A better fix than #4823. tail_logs API is now updated to return an exit code representing the job's status (failed or succeeded). This eliminates the need for a separate API call to fetch the job status. For now, we do not return the job's actual exit c...
2025-02-28T02:19:16Z
If a job fails it should exit with 1 instead of 0 ```bash $ sky launch .. ERROR: Job 439 failed with return code list: [1] ✓ Job finished (status: FAILED). $ echo $? 0 ``` It would be nice if the sky command would exit with 1 instead of 0 to indicate that the job failed 👍
Thanks for the report @kesitrifork. Is this for ease of scripting? As a workaround, would `sky logs [cluster] [job id] --status` work? @concretevitamin Yes exactly, ideally we would switch to using the python interface, but scripting is the quick start. I had not seen that, I thought I had looked, I will try using th...
[ { "body": "```bash\n$ sky launch ..\nERROR: Job 439 failed with return code list: [1]\n✓ Job finished (status: FAILED).\n$ echo $?\n0\n```\n\nIt would be nice if the sky command would exit with 1 instead of 0 to indicate that the job failed 👍", "number": 4599, "title": "If a job fails it should exit wi...
8b466f9d39680e91ea3d73017e19c708db8a5302
{ "head_commit": "935d3c838f5a4add5dfae2e055e49cc7d456c432", "head_commit_message": "minor", "patch_to_review": "diff --git a/sky/backends/cloud_vm_ray_backend.py b/sky/backends/cloud_vm_ray_backend.py\nindex d5c04fa4612..acd92fb3f27 100644\n--- a/sky/backends/cloud_vm_ray_backend.py\n+++ b/sky/backends/cloud_vm_...
[ { "diff_hunk": "@@ -187,6 +191,77 @@ def colored_str(self) -> str:\n return f'{color}{self.value}{colorama.Style.RESET_ALL}'\n \n \n+class JobExitCode(enum.IntEnum):\n+ \"\"\"Job exit code enum.\n+\n+ These codes are used as return codes for job-related operations and as\n+ process exit codes t...
fb048ad8e80598f16c53f89cb6c01254f5558aef
diff --git a/sky/backends/cloud_vm_ray_backend.py b/sky/backends/cloud_vm_ray_backend.py index d5c04fa4612..31a7bfb7eda 100644 --- a/sky/backends/cloud_vm_ray_backend.py +++ b/sky/backends/cloud_vm_ray_backend.py @@ -3823,6 +3823,10 @@ def tail_logs(self, follow: Whether to follow the logs. ta...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-4727@16d1edc
skypilot-org/skypilot
Python
4,727
Unify the advanced config name from `.sky/config` to `.sky/config.yaml` in docs
Fixes #4711
2025-02-16T14:53:01Z
[Docs] Minor: consolidate the naming of advanced user config to `~/.sky/config.yaml` ``` andyl@andylizf-dev-server ~/skypilot (master)> grep -r "~/.sky/config[^.]" . ./sky/clouds/oci.py: in the sky config file, because the ~/.sky/config is not sync-ed to the grep: ./sky/clouds/__pycache__/oci.cpython-311.pyc: binary ...
[ { "body": "```\nandyl@andylizf-dev-server ~/skypilot (master)> grep -r \"~/.sky/config[^.]\" .\n./sky/clouds/oci.py: in the sky config file, because the ~/.sky/config is not sync-ed to the\ngrep: ./sky/clouds/__pycache__/oci.cpython-311.pyc: binary file matches\n./sky/provision/kubernetes/utils.py: \"\"\"A...
9d8d7c6f6bc5482bd10c4c0f548786012c0a9b18
{ "head_commit": "16d1edc74a461ff22891162ceaf8ca3c1093089c", "head_commit_message": "docs: change `.sky/config` to `.sky/config.yaml`", "patch_to_review": "diff --git a/docs/source/reference/kubernetes/kubernetes-ports.rst b/docs/source/reference/kubernetes/kubernetes-ports.rst\nindex 4f8476c1bbc..96a6c8868bc 100...
[ { "diff_hunk": "", "line": null, "original_line": null, "original_start_line": null, "path": "examples/unsloth/unsloth_example.py", "start_line": null, "text": "@user1:\nunsorted?" } ]
1a0718efc2263b5b13c3745b8b50ec6214f1cbd4
diff --git a/docs/source/reference/kubernetes/kubernetes-ports.rst b/docs/source/reference/kubernetes/kubernetes-ports.rst index 4f8476c1bbc..96a6c8868bc 100644 --- a/docs/source/reference/kubernetes/kubernetes-ports.rst +++ b/docs/source/reference/kubernetes/kubernetes-ports.rst @@ -114,7 +114,7 @@ To use this mode: ...
{ "difficulty": "low", "estimated_review_effort": 1, "problem_domain": "Documentation Updates" }
skypilot-org__skypilot-4931@6434ba5
skypilot-org/skypilot
Python
4,931
[API server] attach setup of controllers
<!-- Describe the changes in this PR --> close https://github.com/skypilot-org/skypilot/issues/4783 The total time of first `sky jobs launch` is unchanged. <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> This PR: ![im...
2025-03-12T11:28:12Z
[UX] Show jobs controller setup logs for the first `sky jobs launch` When sky jobs launch for the first time, the output for the controller setup is hidden in `sky logs sky-jobs-controller-xxx 1` which is hard to find and may cause people wonder why a job stay in PENDING for a long time. This is a problem caused by o...
[ { "body": "When sky jobs launch for the first time, the output for the controller setup is hidden in `sky logs sky-jobs-controller-xxx 1` which is hard to find and may cause people wonder why a job stay in PENDING for a long time. \n\nThis is a problem caused by our recent update to API server, where we make th...
849d6fd6294dbc0b5f894c2b351baf569cec60d0
{ "head_commit": "6434ba54561e098393e4eda843bc8cbe3c6e2df6", "head_commit_message": "lint\n\nSigned-off-by: Aylei <rayingecho@gmail.com>", "patch_to_review": "diff --git a/sky/execution.py b/sky/execution.py\nindex 5ff8762acd3..b0261ab0718 100644\n--- a/sky/execution.py\n+++ b/sky/execution.py\n@@ -529,6 +529,13 ...
[ { "diff_hunk": "@@ -529,6 +529,13 @@ def launch(\n ]\n skip_unnecessary_provisioning = True\n \n+ detach_setup = True\n+ if controller_utils.Controllers.from_name(cluster_name) is not None:", "line": null, "original_line": 533, "original_start_line": null, "path": "...
1f781dec9c0c7048d9691c42d8c0482a50d51494
diff --git a/sky/execution.py b/sky/execution.py index 5ff8762acd3..967c79d09d8 100644 --- a/sky/execution.py +++ b/sky/execution.py @@ -529,6 +529,11 @@ def launch( ] skip_unnecessary_provisioning = True + # Attach to setup if the cluster is a controller, so that user can + # see the ...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "New Feature Additions" }
skypilot-org__skypilot-5039@03991d0
skypilot-org/skypilot
Python
5,039
[API server] performance improvement in local mode
<!-- Describe the changes in this PR --> Given same resources, the max parallelism cannot be **exactly** as good as 0.8.0 due to the overhead of uvicorn server and status refresh daemon (see the test result). But this branch can already achieve similar concurrency if user provide `--async` to the CLI (so there is no...
2025-03-26T08:32:28Z
[API server] Local API server has lower concurrency compared to 0.8.0 One of our users runs concurrent `sky.launch` (via Python API) in kubernetes Pod with relatively low resource limit: ``` Requests - CPU: 500m, Memory: 1Gi Limits - CPU: 750m, Memory: 4Gi ``` After upgrading to recent versions with local API server ...
[ { "body": "One of our users runs concurrent `sky.launch` (via Python API) in kubernetes Pod with relatively low resource limit:\n\n```\nRequests - CPU: 500m, Memory: 1Gi\nLimits - CPU: 750m, Memory: 4Gi\n```\n\nAfter upgrading to recent versions with local API server enabled, the max concurrency long-running `s...
1467e7ce9eb45b361e5a46d3ec66ef47acb1f427
{ "head_commit": "03991d0554083dea87d179fe2fc0472f5893e8ee", "head_commit_message": "Address review comments\n\nSigned-off-by: Aylei <rayingecho@gmail.com>", "patch_to_review": "diff --git a/sky/server/requests/executor.py b/sky/server/requests/executor.py\nindex 9ec75f5a68f..18464320d86 100644\n--- a/sky/server/...
[ { "diff_hunk": "@@ -67,10 +68,13 @@ def benchmark_queue_dispatcher(queue_backend: str,\n def enqueue_requests():\n for _ in range(num_requests):\n body = payloads.StatusBody()\n+ body.env_vars = dict({\n+ constants.USER_ID_ENV_VAR: 'dummy',", "line": null, ...
f335e5794cecde14903ff3f5b42b84f9743d154b
diff --git a/sky/server/requests/executor.py b/sky/server/requests/executor.py index 8e3cf1bb7cd..12b927b7459 100644 --- a/sky/server/requests/executor.py +++ b/sky/server/requests/executor.py @@ -18,9 +18,7 @@ See the [README.md](../README.md) for detailed architecture of the executor. """ -import concurrent.futur...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "Performance Optimizations" }
skypilot-org__skypilot-4778@5443f8f
skypilot-org/skypilot
Python
4,778
[API server] allow multiple deployment in one k8s
<!-- Describe the changes in this PR --> close #4777 A quick fix to allow me test multiple API server in one k8s. After this PR, I can deploy an API server under `host:port/test1` and use `sky api login -e http://host:port/test1` to access the server. Minimal Helm command to deploy more API servers after the ...
2025-02-20T15:08:59Z
[API Server] Failed to install multiple API servers in one k8s 1. The chart installs ingress-nginx by default, which conflicts with my existing ingress-nginx. And looks like is no option to disable it. ```bash $ helm template aylei charts/skypilot --namespace skypilot --set ingress.authCredentials=$AUTH_STRING | grep ...
[ { "body": "1. The chart installs ingress-nginx by default, which conflicts with my existing ingress-nginx. And looks like is no option to disable it.\n\n```bash\n$ helm template aylei charts/skypilot --namespace skypilot --set ingress.authCredentials=$AUTH_STRING | grep -A 12 IngressClass\nkind: IngressClass\nm...
6b2b31d8358f3ff8394a7a33ec49e9985ada230f
{ "head_commit": "5443f8f34c60f224d712f1dcf3d7aabb54706798", "head_commit_message": "[API server] allow multiple deployment\n\nSigned-off-by: Aylei <rayingecho@gmail.com>", "patch_to_review": "diff --git a/charts/skypilot/Chart.yaml b/charts/skypilot/Chart.yaml\nindex 743099cacfb..e6580c1c375 100644\n--- a/charts...
[ { "diff_hunk": "@@ -81,12 +81,21 @@ ingress:\n authSecret: null\n # Basic auth credentials in format \"username:encrypted_password\" (only used if ingress.authSecret is not set)\n authCredentials: \"username:$apr1$encrypted_password\"\n- # Specific nodePort to use for the ingress controller\n- # If not ...
2db02e5150474909cc8317c246baf44288f847ca
diff --git a/charts/skypilot/Chart.yaml b/charts/skypilot/Chart.yaml index 743099cacfb..e6580c1c375 100644 --- a/charts/skypilot/Chart.yaml +++ b/charts/skypilot/Chart.yaml @@ -8,3 +8,4 @@ dependencies: - name: ingress-nginx version: 4.11.3 repository: https://kubernetes.github.io/ingress-nginx + condit...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-4802@f24592f
skypilot-org/skypilot
Python
4,802
[core] Lazy import modules to reduce importing time
<!-- Describe the changes in this PR --> Part of #4678 Improvements for the core operations: - sky exec, mean time for 5 runs from 3.45s to 2.747s, about 20% faster - sky launch on existing cluster, mean time for 5 runs from 14.901s to 13.08s, about 12% faster - sky status, mean time for 5 runs from 2.426s to 2...
2025-02-23T08:42:58Z
`import sky` almost imports all subpackages and adds 1s overhead to all operations Found that our backend module is adding a significant startup cost to all operations. Even importing just `sky.utils.common_utils` takes ~0.95s, which is almost the same as importing all of `sky` (~1.04s)! ```bash # Importing all of sky...
I suspect this is in large part due to significant I/O that we do on module load, like setting up databases. > I suspect this is in large part due to significant I/O that we do on module load, like setting up databases. Can we make it lazy? After further investigation, I found that the root cause is our `__init__.py` ...
[ { "body": "Found that our backend module is adding a significant startup cost to all operations. Even importing just `sky.utils.common_utils` takes ~0.95s, which is almost the same as importing all of `sky` (~1.04s)!\n\n```bash\n# Importing all of sky\n$ time python -c \"import sky\"\nreal 0m1.036s\n\n# Just...
90d36ae06177da45a3fe3d872102302a1b880c42
{ "head_commit": "f24592fbfdcf50b68d036165fbd4f9adbbc95c73", "head_commit_message": "remove autouse for the fixture", "patch_to_review": "diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md\nindex e227419c127..397f9f6e1a4 100644\n--- a/CONTRIBUTING.md\n+++ b/CONTRIBUTING.md\n@@ -3,18 +3,19 @@\n Thank you for your inte...
[ { "diff_hunk": "@@ -0,0 +1,55 @@\n+\"\"\"Ensure the lazy import of modules is not reverted.\n+\n+This test ensures that the lazy import of modules is not reverted.\n+\"\"\"\n+import copy\n+import gc\n+import sys\n+\n+import pytest\n+\n+from sky.adaptors.common import LazyImport", "line": null, "original...
9a65e0ad8ea1aaa1013b179850160d46df90a21d
diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md index e227419c127..397f9f6e1a4 100644 --- a/CONTRIBUTING.md +++ b/CONTRIBUTING.md @@ -3,18 +3,19 @@ Thank you for your interest in contributing to SkyPilot! We welcome and value all contributions to the project, including but not limited to: -* [Bug reports](https://gi...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "Performance Optimizations" }
skypilot-org__skypilot-4731@7458e3b
skypilot-org/skypilot
Python
4,731
[Client/Server] fixed server hung during smoke test
<!-- Describe the changes in this PR --> close https://github.com/skypilot-org/skypilot/issues/4732 API server stability enhancements in resource-constrained environments: 1. Optimize memory estimation of API Server - Long worker: 0.25GB => 0.4GB - Short worker: 0.15GB => 0.25GB - For example, the forme...
2025-02-17T15:57:25Z
[Client/Server] server hung on memory starvation # Actual Behavior I launched an API server on an 8c16g VM and set resources limits of the API server container to 7c14g via docker and test the server with parallel requests: ```bash host> docker update --cpus 7 <container-id> host> docker update --memory 14g --memory-...
[ { "body": "# Actual Behavior\n\nI launched an API server on an 8c16g VM and set resources limits of the API server container to 7c14g via docker and test the server with parallel requests:\n\n```bash\nhost> docker update --cpus 7 <container-id>\nhost> docker update --memory 14g --memory-swap 14g <container-id>\...
02abc56a0920a71673c5ef64c5e9e6c00b907c22
{ "head_commit": "7458e3bcdef33b390834304ef3ccd8936cf3628b", "head_commit_message": "relax resource constrain\n\nSigned-off-by: Aylei <rayingecho@gmail.com>", "patch_to_review": "diff --git a/sky/server/requests/executor.py b/sky/server/requests/executor.py\nindex 561cb01466a..d3c801586c2 100644\n--- a/sky/server...
[ { "diff_hunk": "@@ -73,8 +73,13 @@\n # Constants based on profiling the peak memory usage of\n # various sky commands. See `tests/load_test/` for details.\n # Max memory consumption for each request.\n-_PER_BLOCKING_REQUEST_MEM_GB = 0.25\n-_PER_NON_BLOCKING_REQUEST_MEM_GB = 0.15\n+# NOTE(dev): update these cons...
9b2e84f7070115c66663283cbdffbd951e9b86c2
diff --git a/sky/server/common.py b/sky/server/common.py index df5bd676c8a..412e5ac0eb7 100644 --- a/sky/server/common.py +++ b/sky/server/common.py @@ -15,7 +15,6 @@ import colorama import filelock -import psutil import pydantic import requests @@ -148,13 +147,14 @@ def get_api_server_status(endpoint: Optional...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "Performance Optimizations" }
skypilot-org__skypilot-4592@d0798a9
skypilot-org/skypilot
Python
4,592
Change added `USER_HASH` in models name to full digits (8) instead of half (4) to prevent potential hash conflict
<!-- Describe the changes in this PR --> Resolves #4143 <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the relevant ones): - [x] Code formatting: `bash format.sh` - [ ] Any manual or new tests for this PR (...
2025-01-18T13:30:13Z
[BUG][AWS] Serving a new controller can result in terminating previously served models. When these conditions are true, a new controller may terminate existing resources being served by another controller. * User id hash matches last 4 characters, but has a difference otherwise. * Service name is equivalent This...
Hi @JGSweets ! Thanks for reporting this error. Just want to make sure, this is for multiple users (in multiple laptops) running SkyPilot in a shared AWS project? > Hi @JGSweets ! Thanks for reporting this error. Just want to make sure, this is for multiple users (in multiple laptops) running SkyPilot in a shared AWS p...
[ { "body": "When these conditions are true, a new controller may terminate existing resources being served by another controller.\r\n* User id hash matches last 4 characters, but has a difference otherwise.\r\n* Service name is equivalent\r\n\r\n\r\nThis may results in a model node's `cluster_name` with the same...
b142e0b03d7e3ea94079ca0000645fb63e873cdc
{ "head_commit": "d0798a9874f9629e85520ce1d53e40a1fe0c21d8", "head_commit_message": "style: format and type hints", "patch_to_review": "diff --git a/sky/backends/backend_utils.py b/sky/backends/backend_utils.py\nindex 7385c7cb9fc..28e6d1ce0f0 100644\n--- a/sky/backends/backend_utils.py\n+++ b/sky/backends/backend...
[ { "diff_hunk": "@@ -98,10 +97,12 @@ def _is_valid_user_hash(user_hash: Optional[str]) -> bool:\n return user_hash\n \n hash_str = user_and_hostname_hash()\n- user_hash = hashlib.md5(hash_str.encode()).hexdigest()[:USER_HASH_LENGTH]\n- if not _is_valid_user_hash(user_hash):\n+ hex_hash =...
b5739a24edd9929b3e8c2f82d3b34816daea80af
diff --git a/sky/backends/backend_utils.py b/sky/backends/backend_utils.py index 7385c7cb9fc..28e6d1ce0f0 100644 --- a/sky/backends/backend_utils.py +++ b/sky/backends/backend_utils.py @@ -997,7 +997,7 @@ def write_cluster_config( # Read the cluster name from the tmp yaml file, to take the backward # compat...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-4587@ed5870b
skypilot-org/skypilot
Python
4,587
[k8s] Show enabled contexts in `sky check`
Closes #4570. Some examples: ### WIthout `allowed_contexts` set <img width="987" alt="image" src="https://github.com/user-attachments/assets/6cc24c42-4b1d-4a96-a5f0-fed21a082d14" /> ### With `allowed_contexts` set <img width="989" alt="image" src="https://github.com/user-attachments/assets/7dc5464d-462c-...
2025-01-18T00:01:45Z
[k8s] Show enabled contexts (allowed_contexts) after `sky check` At the end of `sky check` we should show the context(s) that will be used for the Kubernetes cluster. Useful for users to know which ones are enabled and disabled, instead of having to guess when specifying with `--regions`. ``` 🎉 Enabled clouds 🎉 ...
[ { "body": "At the end of `sky check` we should show the context(s) that will be used for the Kubernetes cluster. Useful for users to know which ones are enabled and disabled, instead of having to guess when specifying with `--regions`.\r\n\r\n```\r\n🎉 Enabled clouds 🎉\r\n ✔ AWS\r\n ✔ Kubernetes: context1, c...
6b23582d9f2285af2816999185947b6ce0df2c81
{ "head_commit": "ed5870bfadb08e7eaa9d53526257f04a2268ff3f", "head_commit_message": "Show context in check", "patch_to_review": "diff --git a/sky/check.py b/sky/check.py\nindex 1ab92cb1af6..d2877fd9c0e 100644\n--- a/sky/check.py\n+++ b/sky/check.py\n@@ -155,7 +155,8 @@ def get_all_clouds():\n # Pretty pri...
[ { "diff_hunk": "@@ -222,3 +223,32 @@ def get_cloud_credential_file_mounts(\n r2_credential_mounts = cloudflare.get_credential_file_mounts()\n file_mounts.update(r2_credential_mounts)\n return file_mounts\n+\n+\n+def _format_enabled_cloud(cloud_name: str) -> str:\n+ if cloud_name == repr(s...
50168639602abf9ebe7ce8ec163ec146220f3789
diff --git a/sky/check.py b/sky/check.py index 1ab92cb1af6..f32e4985079 100644 --- a/sky/check.py +++ b/sky/check.py @@ -155,7 +155,8 @@ def get_all_clouds(): # Pretty print for UX. if not quiet: enabled_clouds_str = '\n :heavy_check_mark: '.join( - [''] + sorted(all_enabl...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "New Feature Additions" }
skypilot-org__skypilot-4584@12631c9
skypilot-org/skypilot
Python
4,584
[SKY-1304] AWS throws `InvalidGroup.Duplicate` when user submits a lot of parallel launch
<!-- Describe the changes in this PR --> Fix SKY 1304 Fixes #4578 Root cause: - [Get_security_groups_from_vpc_ids](get_security_groups_from_vpc_ids) can't query security groups when security groups is under creating. Solution: - When `InvalidGroup.Duplicate` exception occurs during security groups creating, que...
2025-01-17T09:59:13Z
[AWS] AWS throws `InvalidGroup.Duplicate` when user submits a lot of parallel launch To reproduce: `seq 10 | xargs -P 16 -n 1 bash -c 'sky launch -c test-$0 --cpus 2 --use-spot --down -i0 -y &'` Observed this with error ``` ot error(s) in AWS:[sky.exceptions.NoClusterLaunchedError] Failed to create security group. Err...
Does this need to be a new VPC that you haven't used skypilot on before? I don't think it is a new VPC. The issue happens for only some of the launches. This issue happens for some of the tests with `pytest tests/test_smoke.py --aws`.
[ { "body": "To reproduce:\n`seq 10 | xargs -P 16 -n 1 bash -c 'sky launch -c test-$0 --cpus 2 --use-spot --down -i0 -y &'`\n\nObserved this with error\n```\not error(s) in AWS:[sky.exceptions.NoClusterLaunchedError] Failed to create security group. Error: botocore.exceptions.ClientError: An error occurred (Inval...
11861fd35820ff0db76ecce1dc9a644db4ffb8f7
{ "head_commit": "12631c90a164808fdf3fe162fbe0edac7f630d8a", "head_commit_message": "wip", "patch_to_review": "diff --git a/sky/provision/aws/config.py b/sky/provision/aws/config.py\nindex ffa87c3a011..d07d364c56d 100644\n--- a/sky/provision/aws/config.py\n+++ b/sky/provision/aws/config.py\n@@ -543,19 +543,41 @@ ...
[ { "diff_hunk": "@@ -543,19 +543,41 @@ def _configure_security_group(ec2, vpc_id: str, expected_sg_name: str,\n return sg_ids\n \n \n-def _get_or_create_vpc_security_group(ec2, vpc_id: str,\n- expected_sg_name: str) -> Any:\n- # Figure out which security groups with th...
9108e5ba0d42fb6888b4a513fddb6718039bb611
diff --git a/sky/provision/aws/config.py b/sky/provision/aws/config.py index acc6fcb0e56..a878bbe65c7 100644 --- a/sky/provision/aws/config.py +++ b/sky/provision/aws/config.py @@ -553,17 +553,28 @@ def _configure_security_group(ec2, vpc_id: str, expected_sg_name: str, def _get_or_create_vpc_security_group(ec2, vpc_...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-4430@3818b1b
skypilot-org/skypilot
Python
4,430
[Core] Execute setup when `--detach-setup` and no `run` section
<!-- Describe the changes in this PR --> Fixes #4419 <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the relevant ones): - [ ] Code formatting: `bash format.sh` - [ ] Any manual or new tests for this PR (p...
2024-12-03T00:36:58Z
[Core] `setup` will not run if you don't have a `run` section and `--detach-setup` is specified <!-- Describe the bug report / feature request here --> <!-- If relevant, fill in versioning info to help us troubleshoot --> _Version & Commit info:_ * `sky -v`: PLEASE_FILL_IN * `sky -c`: PLEASE_FILL_IN
To reproduce: ``` # test.yaml setup: hostname ``` ``` sky launch -c test-detach --detach-setup test.yaml ``` The setup is not executed, due to: https://github.com/skypilot-org/skypilot/blob/23f9821bd165c8d154a44c3505a18b1e28cc0f48/sky/backends/cloud_vm_ray_backend.py#L3463-L3465
[ { "body": "<!-- Describe the bug report / feature request here -->\r\n\r\n\r\n\r\n\r\n<!-- If relevant, fill in versioning info to help us troubleshoot -->\r\n_Version & Commit info:_\r\n* `sky -v`: PLEASE_FILL_IN\r\n* `sky -c`: PLEASE_FILL_IN\r\n", "number": 4419, "title": "[Core] `setup` will not run ...
68723df97c7c981887ba9100c510aca953f45c11
{ "head_commit": "3818b1b4fab31299dda7f63fb41097b6c6fa493b", "head_commit_message": "Execute setup when --detach-setup and no run section", "patch_to_review": "diff --git a/sky/backends/cloud_vm_ray_backend.py b/sky/backends/cloud_vm_ray_backend.py\nindex 5682cf24586..19e76daa082 100644\n--- a/sky/backends/cloud_...
[ { "diff_hunk": "@@ -3460,15 +3462,23 @@ def _execute(\n Returns:\n Job id if the task is submitted to the cluster, None otherwise.\n \"\"\"\n- if task.run is None:\n+ if task.run is None and self._setup_cmd is None:\n logger.info('Run commands not specified ...
5824d2410960f3a26f8eaa74aca4fb601a2d123a
diff --git a/sky/__init__.py b/sky/__init__.py index b851775dabf..4e720d63ce0 100644 --- a/sky/__init__.py +++ b/sky/__init__.py @@ -105,6 +105,7 @@ def set_proxy_env_var(proxy_var: str, urllib_var: Optional[str]): from sky.data import StoreType from sky.execution import exec # pylint: disable=redefined-builtin fro...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-4264@d738724
skypilot-org/skypilot
Python
4,264
[Core] Avoid PENDING job to be set to FAILED and speed up job scheduling
<!-- Describe the changes in this PR --> Fixes #4263 It also speed up the job scheduling by avoid update status for all the jobs: Previously, gap between jobs to be scheduled is ~4-5s, ie with 60s duration of a task, there will be 12 jobs running in parallel (the additional RUNNING (with more than 1 min job dur...
2024-11-05T19:44:50Z
[Core] Submitting 1000 jobs to a cluster <!-- Describe the bug report / feature request here --> `sky launch -c test-many-jobs --cloud aws --cpus 16 --region us-east-1` and submit jobs: ``` import subprocess from multiprocessing.pool import ThreadPool def run_task(task): print(f'Running task {task}') ...
https://gist.github.com/Michaelvll/9325ff98e6bd2b454dfd661fb9c0ebab
[ { "body": "<!-- Describe the bug report / feature request here -->\r\n\r\n\r\n`sky launch -c test-many-jobs --cloud aws --cpus 16 --region us-east-1`\r\nand submit jobs:\r\n```\r\nimport subprocess\r\nfrom multiprocessing.pool import ThreadPool\r\n\r\ndef run_task(task):\r\n print(f'Running task {task}')\r\n...
1dcd2f009c1989cbc130e5b8490170a5a96c3e23
{ "head_commit": "d73872412cbeb39293c77466c53de4886e5803b9", "head_commit_message": "Merge branch 'master' of github.com:skypilot-org/skypilot into longer-wait-for-pending-to-failed", "patch_to_review": "diff --git a/examples/multi_echo.py b/examples/multi_echo.py\nindex a9fe49fad2d..7f310a4bb6b 100644\n--- a/exa...
[ { "diff_hunk": "@@ -537,32 +542,44 @@ def update_job_status(job_ids: List[int],\n if len(job_ids) == 0:\n return []\n \n- # TODO: if too slow, directly query against redis.\n ray_job_ids = [make_ray_job_id(job_id) for job_id in job_ids]\n-\n job_client = _create_ray_job_submission_client(...
da08bb3e9b28531aea143a2273218bb363e15c0d
diff --git a/examples/multi_echo.py b/examples/multi_echo.py index a9fe49fad2d..7f310a4bb6b 100644 --- a/examples/multi_echo.py +++ b/examples/multi_echo.py @@ -31,13 +31,13 @@ def run(cluster: Optional[str] = None, cloud: Optional[str] = None): # Submit multiple tasks in parallel to trigger queueing behaviors. ...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-4881@1829dfb
skypilot-org/skypilot
Python
4,881
[API server] better error message on API version mismatch
<!-- Describe the changes in this PR --> close #4878 Tested locally by modifying the version constant. UX screenshots: ## Remote server too old <img width="1036" alt="image" src="https://github.com/user-attachments/assets/0eeaf7e1-bf84-4507-90ed-841f2be3ecc9" /> ## Client too old (remote server) <img ...
2025-03-05T08:34:50Z
[API server] Refine error message of version mismatch One of our users reported the following error when trying to use a remote API server: ``` RuntimeError: SkyPilot API server is too old: v1 (client version is v3). Please restart the SkyPilot API server with: sky api stop; sky api start ``` This is only valid for l...
[ { "body": "One of our users reported the following error when trying to use a remote API server:\n\n```\nRuntimeError: SkyPilot API server is too old: v1 (client version is v3). Please restart the SkyPilot API server with: sky api stop; sky api start\n```\n\nThis is only valid for local API servers and should b...
8a1a6c7334d470fcfddb3455edf6d7eb2a24fa0c
{ "head_commit": "1829dfbfa2b881f318b9b3f93b11bd91ab7b9151", "head_commit_message": "Address review comments\n\nSigned-off-by: Aylei <rayingecho@gmail.com>", "patch_to_review": "diff --git a/sky/server/common.py b/sky/server/common.py\nindex 1997ddf8c00..0e088365181 100644\n--- a/sky/server/common.py\n+++ b/sky/s...
[ { "diff_hunk": "@@ -46,12 +46,29 @@\n API_SERVER_CLIENT_DIR = pathlib.Path('~/.sky/api_server/clients')\n RETRY_COUNT_ON_TIMEOUT = 3\n \n-SKY_API_VERSION_WARNING = (\n+SKY_CLIENT_TOO_OLD_WARNING = (\n+ f'{colorama.Fore.YELLOW}Your SkyPilot client is too old: '\n+ f'v{{client_version}} (server version is v...
74910edfdbd6dae6af3d3bcaf882e4b90f336d3e
diff --git a/sky/exceptions.py b/sky/exceptions.py index 3239ba51e2f..c4712371ca9 100644 --- a/sky/exceptions.py +++ b/sky/exceptions.py @@ -477,6 +477,11 @@ def __init__(self, server_url: str): f'Try: curl {server_url}/api/health') +class APIVersionMismatchError(RuntimeError): + """Raised when the ...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-4692@1cf4d5c
skypilot-org/skypilot
Python
4,692
[UX] Auto-exclude unavailable kubernetes contexts
<!-- Describe the changes in this PR --> close https://github.com/skypilot-org/skypilot/issues/2807 - ~~Excludes unavailable kubernetes contexts at runtime~~ (**UPDATED**: see https://github.com/skypilot-org/skypilot/pull/4692#discussion_r1952092329); - skip unavailable kubernetes contexts, the context will be...
2025-02-11T12:18:20Z
[k8s][ux] Auto-exclude stale Kubernetes cloud I often terminate a Kubernetes cluster externally using the cloud console/cli (e.g., `gcloud container clusters delete <cluster-name> --region us-central1-c`), but I forget to run sky check to update the list of enabled clouds. As a result, the next `sky launch` fails: ...
This is also related to #3013 Going to self-assign and work on this! This issue is stale because it has been open 120 days with no activity. Remove stale label or comment or this will be closed in 10 days. This issue is stale because it has been open 120 days with no activity. Remove stale label or comment or this wil...
[ { "body": "I often terminate a Kubernetes cluster externally using the cloud console/cli (e.g., `gcloud container clusters delete <cluster-name> --region us-central1-c`), but I forget to run sky check to update the list of enabled clouds.\r\n\r\nAs a result, the next `sky launch` fails:\r\n```\r\nsky.exceptions...
bba4f6f63355a8872d2eaa2f492a56c6a52bf9ca
{ "head_commit": "1cf4d5ccd6941b3f5bd08b3473db04e88862f656", "head_commit_message": "address review comments\n\nSigned-off-by: Aylei <rayingecho@gmail.com>", "patch_to_review": "diff --git a/sky/clouds/kubernetes.py b/sky/clouds/kubernetes.py\nindex 3bab7687f48..7c44e79f1b2 100644\n--- a/sky/clouds/kubernetes.py\...
[ { "diff_hunk": "@@ -131,6 +135,17 @@ def wrapper(*args, **kwargs):\n backoff = common_utils.Backoff(initial_backoff=retry_interval,\n max_backoff_factor=3)\n \n+ # try to capture the kubernetes context from parameters\n+ if context_par...
9e963c58abe1138350ec08314908b6a78de63509
diff --git a/sky/clouds/kubernetes.py b/sky/clouds/kubernetes.py index 239a3694c11..5de47f89d65 100644 --- a/sky/clouds/kubernetes.py +++ b/sky/clouds/kubernetes.py @@ -2,9 +2,10 @@ import os import re import typing -from typing import Dict, Iterator, List, Optional, Tuple, Union +from typing import Dict, Iterator, ...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "New Feature Additions" }
skypilot-org__skypilot-4194@0a12b96
skypilot-org/skypilot
Python
4,194
[Jobs] Fix `is_chain` to check in- and out-degrees
<!-- Describe the changes in this PR --> Resolve #4187 - Adds `in_degree` checks to ensure the DAG has only one start and one end node. - Ensures all nodes have at most one incoming and one outgoing edge. <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are test...
2024-10-26T22:48:50Z
Bug: `is_chain` function inaccurately detects chain structure <!-- Describe the bug report / feature request here --> The `is_chain` function fails to accurately detect chain structures. The current logic only checks that there's one node with an out-degree of 0 and all other nodes have an out-degree of 1. However, ...
@cblmemo PTAL, thanks!
[ { "body": "<!-- Describe the bug report / feature request here -->\r\n\r\nThe `is_chain` function fails to accurately detect chain structures. The current logic only checks that there's one node with an out-degree of 0 and all other nodes have an out-degree of 1. However, this overlooks cases where multiple nod...
bb39707526c926619af6b97cdebedbbd3193ea5b
{ "head_commit": "0a12b96ab822988cc3d95c6d01aee3d6a261291e", "head_commit_message": "fix: is_chain also checks out_degrees", "patch_to_review": "diff --git a/sky/dag.py b/sky/dag.py\nindex 4af5adc76b5..4d8f585d8f8 100644\n--- a/sky/dag.py\n+++ b/sky/dag.py\n@@ -55,21 +55,21 @@ def get_graph(self):\n retur...
[ { "diff_hunk": "@@ -55,21 +55,21 @@ def get_graph(self):\n return self.graph\n \n def is_chain(self) -> bool:\n- # NOTE: this method assumes that the graph has no cycle.\n- is_chain = True\n- visited_zero_out_degree = False\n- for node in self.graph.nodes:\n- o...
bfb300fd9b050a98f2fe4b76c8f8a71ec353e145
diff --git a/sky/dag.py b/sky/dag.py index 4af5adc76b5..0d339366831 100644 --- a/sky/dag.py +++ b/sky/dag.py @@ -55,21 +55,25 @@ def get_graph(self): return self.graph def is_chain(self) -> bool: - # NOTE: this method assumes that the graph has no cycle. - is_chain = True - visited_...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-4165@56edbb8
skypilot-org/skypilot
Python
4,165
Upload all cloud credentials to sky cluster regardless of sky check
<!-- Describe the changes in this PR --> Resolve https://github.com/skypilot-org/skypilot/issues/3977 <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the relevant ones): - [x] Code formatting: `bash format.sh...
2024-10-24T21:16:34Z
[Core] Upload cloud credentials for bucket access even when cloud is disabled in SkyPilot <!-- Describe the bug report / feature request here --> A user reported that their gcp credential may not have the full permission to let `sky check` enable the cloud, but when they launch clusters in kubernetes, they still wan...
[ { "body": "<!-- Describe the bug report / feature request here -->\r\n\r\nA user reported that their gcp credential may not have the full permission to let `sky check` enable the cloud, but when they launch clusters in kubernetes, they still want the credential to be uploaded for the access to GCS bucket.\r\n\r...
6dc386bab615bdb9fd83dc10722947de7c2056c5
{ "head_commit": "56edbb874806f33b63fa246f21d01a6f242cc9f6", "head_commit_message": "Upload all cloud credentials to sky cluster regardless of sky check", "patch_to_review": "diff --git a/sky/check.py b/sky/check.py\nindex 9ac2848733c..1854d6383f5 100644\n--- a/sky/check.py\n+++ b/sky/check.py\n@@ -1,4 +1,5 @@\n ...
[ { "diff_hunk": "@@ -194,19 +195,25 @@ def get_cached_enabled_clouds_or_refresh(\n def get_cloud_credential_file_mounts(\n excluded_clouds: Optional[Iterable[sky_clouds.Cloud]]\n ) -> Dict[str, str]:\n- \"\"\"Returns the files necessary to access all enabled clouds.\n+ \"\"\"Returns the files neces...
ef5eefcfafc36739b1247592667fbe274053e03a
diff --git a/sky/check.py b/sky/check.py index 9ac2848733c..dcaa349d234 100644 --- a/sky/check.py +++ b/sky/check.py @@ -1,4 +1,5 @@ """Credential checks: check cloud credentials and enable clouds.""" +import os import traceback from types import ModuleType from typing import Dict, Iterable, List, Optional, Tuple, ...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "New Feature Additions" }
skypilot-org__skypilot-4454@1d67da7
skypilot-org/skypilot
Python
4,454
[Storage] Continue storage deletion in `sky storage delete` when some fail
<!-- Describe the changes in this PR --> Fix #4050. Make `sky storage delete` continue deleting remaining storages when some fail, instead of aborting the entire operation. Failed deletions are reported individually. We Implemented this feature via adding a `continue_on_error` parameter in `run_in_parallel` <!-- De...
2024-12-10T21:45:02Z
[Storage] `sky storage delete -a` aborted when deletion of one storage failed <!-- Describe the bug report / feature request here --> `sky storage delete -a` aborted for all storage when deletion of one storage failed. Expected behaviour: print out error for corresponding storage while still delete all other storage...
[ { "body": "<!-- Describe the bug report / feature request here -->\r\n\r\n`sky storage delete -a` aborted for all storage when deletion of one storage failed. Expected behaviour: print out error for corresponding storage while still delete all other storages.\r\n\r\n```bash\r\nsky storage delete -a\r\nDeleting ...
d9bb51a66a33ed06f9f0c791ac3ed5ad93ffa587
{ "head_commit": "1d67da7ed0fbd0c7113ba268595ae2c589b24a43", "head_commit_message": "chore: remove TODO", "patch_to_review": "diff --git a/sky/cli.py b/sky/cli.py\nindex 1faf0003ff9..afbcec5e13e 100644\n--- a/sky/cli.py\n+++ b/sky/cli.py\n@@ -3525,11 +3525,10 @@ def storage_delete(names: List[str], all: bool, yes...
[ { "diff_hunk": "@@ -3543,7 +3542,18 @@ def storage_delete(names: List[str], all: bool, yes: bool): # pylint: disable=r\n abort=True,\n show_default=True)\n \n- subprocess_utils.run_in_parallel(sky.storage_delete, names)\n+ def delete_storage(name: str) -> None:\n+ t...
a2e9b05531ba16ff385b16d4b3217b275ac40c82
diff --git a/sky/cli.py b/sky/cli.py index 27948f9ec85..34793a74995 100644 --- a/sky/cli.py +++ b/sky/cli.py @@ -3530,11 +3530,11 @@ def storage_delete(names: List[str], all: bool, yes: bool): # pylint: disable=r if sum([bool(names), all]) != 1: raise click.UsageError('Either --all or a name must be spec...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-3996@a5f1a31
skypilot-org/skypilot
Python
3,996
[Docs] Fix inconsistent example in spot instance documentation
Fixes #3995
2024-09-25T23:11:40Z
[Docs] Inconsistent Example in Spot Instance Documentation The documentation states that when one spot replica is available, SkyServe will provision an on-demand replica to meet the target number of replicas. However, the accompanying console output shows a scenario where there are 0 spot replicas available, which cont...
[ { "body": "The documentation states that when one spot replica is available, SkyServe will provision an on-demand replica to meet the target number of replicas. However, the accompanying console output shows a scenario where there are 0 spot replicas available, which contradicts this explanation and could confu...
82aa0c3d9cc8cbf6039c18d9e5b01a9256040a33
{ "head_commit": "a5f1a31311173d3fefa1f23134428f6dc21baf74", "head_commit_message": "[Docs] Fix inconsistent example in spot instance documentation\n\nFixes #3995", "patch_to_review": "diff --git a/docs/source/serving/spot-policy.rst b/docs/source/serving/spot-policy.rst\nindex ff23b328705..1b812fea605 100644\n--...
[ { "diff_hunk": "@@ -96,7 +96,7 @@ When the service is up, we can check the status of the service and the replicas\n http-server 3 1 - 1 mins ago 1x GCP(vCPU=2) PROVISIONING us-east1 \n http-server 4 1 - 1 min ago 1x G...
010b5c8721ee30a69b3e6c9d15583450fa0aa9a2
diff --git a/docs/source/serving/spot-policy.rst b/docs/source/serving/spot-policy.rst index ff23b328705..f9785d0eeb0 100644 --- a/docs/source/serving/spot-policy.rst +++ b/docs/source/serving/spot-policy.rst @@ -96,7 +96,7 @@ When the service is up, we can check the status of the service and the replicas http-ser...
{ "difficulty": "low", "estimated_review_effort": 1, "problem_domain": "Documentation Updates" }
skypilot-org__skypilot-4045@df90537
skypilot-org/skypilot
Python
4,045
[K8s] Add user hash to the kind config for multi-user system permission issue
<!-- Describe the changes in this PR --> Fixes #3533 <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the relevant ones): - [ ] Code formatting: `bash format.sh` - [ ] Any manual or new tests for this PR (ple...
2024-10-07T17:57:00Z
[K8S] `sky local up` failed on a multi-user system <!-- Describe the bug report / feature request here --> On commit `7f30ce5e` Currently the `sky local up` will create a config yaml under `/tmp` and does not clean it up after local cluster creation. If another user on the same Linux OS tries to run `sky local up...
We can probably attach a user hash to the yaml name : )
[ { "body": "<!-- Describe the bug report / feature request here -->\r\n\r\nOn commit `7f30ce5e`\r\n\r\nCurrently the `sky local up` will create a config yaml under `/tmp` and does not clean it up after local cluster creation. If another user on the same Linux OS tries to run `sky local up`, it will show no permi...
d5b6d89c83ea1ee7258f68314da4c6f8add83e04
{ "head_commit": "df90537affe184611e30221188c644c58e74ef3a", "head_commit_message": "fix", "patch_to_review": "diff --git a/sky/cli.py b/sky/cli.py\nindex c538c99aeb3..b9756b63083 100644\n--- a/sky/cli.py\n+++ b/sky/cli.py\n@@ -5097,7 +5097,8 @@ def _deploy_local_cluster(gpus: bool):\n \n # Get directory of s...
[ { "diff_hunk": "@@ -5097,7 +5097,8 @@ def _deploy_local_cluster(gpus: bool):\n \n # Get directory of script and run it from there\n cwd = os.path.dirname(os.path.abspath(up_script_path))\n- run_command = up_script_path + ' --gpus' if gpus else up_script_path\n+ run_command = up_script_path + f' --...
1bd4de46b899d0aae1fb70d638154fd4ae0d67e4
diff --git a/sky/cli.py b/sky/cli.py index c538c99aeb3..093db23adbf 100644 --- a/sky/cli.py +++ b/sky/cli.py @@ -5097,7 +5097,8 @@ def _deploy_local_cluster(gpus: bool): # Get directory of script and run it from there cwd = os.path.dirname(os.path.abspath(up_script_path)) - run_command = up_script_path +...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-3792@8bb0647
skypilot-org/skypilot
Python
3,792
Allow setting the workdir using environment variables
<!-- Describe the changes in this PR --> Allows the workdir to be configured using environment variables. Closes #3271 <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the relevant ones): - [x] Code format...
2024-07-27T04:52:02Z
Allow setting the `workdir` using environment variables For situations when the working directory is not known in advance (e.g., on remote machines), it would be helpful to be able to dynamically populate the `workdir` value using an environment variable using the following syntax: ```yaml envs: MY_WORKDIR: "som...
This issue is stale because it has been open 120 days with no activity. Remove stale label or comment or this will be closed in 10 days. This issue was closed because it has been stalled for 10 days with no activity.
[ { "body": "For situations when the working directory is not known in advance (e.g., on remote machines), it would be helpful to be able to dynamically populate the `workdir` value using an environment variable using the following syntax:\r\n\r\n```yaml\r\nenvs:\r\n MY_WORKDIR: \"some/path/\"\r\n\r\nworkfir: ${...
e794971548584cdc9c8d6c1bde4b734c151ef3c4
{ "head_commit": "8bb0647b69739e05bd2aff5c43b841b146f53b63", "head_commit_message": "Format", "patch_to_review": "diff --git a/sky/task.py b/sky/task.py\nindex cf26e13717a..cebc616dc6d 100644\n--- a/sky/task.py\n+++ b/sky/task.py\n@@ -393,6 +393,11 @@ def from_yaml_config(\n config['service'] = _fill_...
[ { "diff_hunk": "@@ -146,3 +146,14 @@ def test_invalid_empty_envs(tmp_path):\n with pytest.raises(ValueError) as e:\n Task.from_yaml(config_path)\n assert 'Environment variable \\'env_key2\\' is None.' in e.value.args[0]\n+\n+\n+def test_replace_envs_in_path(tmpdir, tmp_path):", "line": null,...
c8a9a50db735e1e58b32fa5d68ac1343302fbe6a
diff --git a/sky/task.py b/sky/task.py index cf26e13717a..cebc616dc6d 100644 --- a/sky/task.py +++ b/sky/task.py @@ -393,6 +393,11 @@ def from_yaml_config( config['service'] = _fill_in_env_vars(config['service'], config.get('envs', {})) + # Fill i...
{ "difficulty": "low", "estimated_review_effort": 2, "problem_domain": "New Feature Additions" }
skypilot-org__skypilot-3786@5526910
skypilot-org/skypilot
Python
3,786
[k8s] Fix recursive decoration of kubernetes API methods
Closes #3774. This was a regression in https://github.com/skypilot-org/skypilot/pull/3674 which decorated the kubernetes API methods multiple times, blowing up the stack depth and causing the recursion error. This PR fixes it by decorating methods only once. Tested (run the relevant ones): - [x] Code formattin...
2024-07-25T23:29:11Z
[k8s][Serve] Replica prober failing with RecursionError on k8s Repro: `pytest tests/test_smoke.py::test_skyserve_update_autoscale --cloud kubernetes` stays stuck. When inspected with `sky serve logs t-ss-update-auto-7q-4975-e2 --controller`: ``` I 07-23 05:22:14 httptools_impl.py:466] 127.0.0.1:51340 - "POST /co...
Oh look at that, I'm seeing this too FYI: I switched down to commit aea732214ed7997b8c15f93f9b764f9175a6d5db and this problem isn't triggering Nevermind, it does start triggering at some point. Possibly related to https://github.com/skypilot-org/skypilot/pull/3674. Reverting that change fixes this. Looks like adding th...
[ { "body": "Repro: `pytest tests/test_smoke.py::test_skyserve_update_autoscale --cloud kubernetes` stays stuck.\r\n\r\nWhen inspected with `sky serve logs t-ss-update-auto-7q-4975-e2 --controller`:\r\n\r\n```\r\nI 07-23 05:22:14 httptools_impl.py:466] 127.0.0.1:51340 - \"POST /controller/load_balancer_sync HTTP/...
eb03d8f4af77f5496a40322a62fe0ff78a870225
{ "head_commit": "55269105fbfd966ff5eb76ee4c870d3913c57f23", "head_commit_message": "lint", "patch_to_review": "diff --git a/sky/adaptors/kubernetes.py b/sky/adaptors/kubernetes.py\nindex 7f52a099f56..e75149836ae 100644\n--- a/sky/adaptors/kubernetes.py\n+++ b/sky/adaptors/kubernetes.py\n@@ -33,8 +33,11 @@\n def ...
[ { "diff_hunk": "@@ -33,8 +33,11 @@\n def _decorate_methods(obj, decorator):\n for attr_name in dir(obj):\n attr = getattr(obj, attr_name)\n- if callable(attr) and not attr_name.startswith('__'):\n- setattr(obj, attr_name, decorator(attr))\n+ if callable(attr) and not attr_na...
c98fc485df92d74bd4830ce24a642cc62e15ee20
diff --git a/sky/adaptors/kubernetes.py b/sky/adaptors/kubernetes.py index 7f52a099f56..52eb339d213 100644 --- a/sky/adaptors/kubernetes.py +++ b/sky/adaptors/kubernetes.py @@ -4,6 +4,7 @@ import logging import os +from typing import Any, Callable, Set from sky.adaptors import common from sky.sky_logging import...
{ "difficulty": "low", "estimated_review_effort": 2, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-3784@d8cb16b
skypilot-org/skypilot
Python
3,784
[Tests] Fix test_docker_storage_mounts for azure blob
Closes #3781. Tested: `pytest tests/test_smoke.py::test_docker_storage_mounts --azure`
2024-07-25T05:34:10Z
[Azure] `test_docker_storage_mounts` needs to be updated for azure blob storage `pytest tests/test_smoke.py::test_docker_storage_mounts --azure` fails with: ``` + aws s3 ls sky-test-17217151744714668/hello.txt || gsutil ls gs://sky-test-17217151744714668/hello.txt An error occurred (NoSuchBucket) when calling th...
[ { "body": "`pytest tests/test_smoke.py::test_docker_storage_mounts --azure` fails with:\r\n\r\n```\r\n+ aws s3 ls sky-test-17217151744714668/hello.txt || gsutil ls gs://sky-test-17217151744714668/hello.txt\r\n\r\nAn error occurred (NoSuchBucket) when calling the ListObjectsV2 operation: The specified bucket doe...
b02a66de1ecb3c36b8d0dce7a463f6a74f85d46b
{ "head_commit": "d8cb16ba7b92393b720e05eb7f06cf9f280e535c", "head_commit_message": "lint", "patch_to_review": "diff --git a/tests/test_smoke.py b/tests/test_smoke.py\nindex 325a836cf4c..ec431895ca0 100644\n--- a/tests/test_smoke.py\n+++ b/tests/test_smoke.py\n@@ -47,6 +47,7 @@\n from sky import global_user_state...
[ { "diff_hunk": "@@ -1189,9 +1190,17 @@ def test_docker_storage_mounts(generic_cloud: str, image_id: str):\n template = jinja2.Template(template_str)\n # ubuntu 18.04 does not support fuse3, and blobfuse2 depends on fuse3.\n azure_mount_unsupported_ubuntu_version = '18.04'\n+ # Commands to verify ...
8eec55fd2a9e6f8c33843e063d9494779f01e4b1
diff --git a/tests/test_smoke.py b/tests/test_smoke.py index 325a836cf4c..3647be5a499 100644 --- a/tests/test_smoke.py +++ b/tests/test_smoke.py @@ -47,6 +47,7 @@ from sky import global_user_state from sky import jobs from sky import serve +from sky import skypilot_config from sky.adaptors import cloudflare from s...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Test Suite / CI Enhancements" }
skypilot-org__skypilot-4067@ff528a5
skypilot-org/skypilot
Python
4,067
DAG API Enhancements: Introducing Downstream Task Parsing and Explicit Flow Definition
<!-- Describe the changes in this PR --> Resolve #4054 This PR introduces the `dependencies` feature to SkyPilot, enabling users to specify task dependencies in their YAML configurations. This enhancement allows for more complex workflow definitions and ensures proper execution order of interdependent tasks. ...
2024-10-10T23:31:30Z
[Jobs] API Design for DAG execution <!-- Describe the bug report / feature request here --> We should think about the API to specify an DAG of tasks. Several choices: 1. Add an attribute to `sky.Task` for predecessor/successor 2. Add a dedicated yaml file for node dependencies 3. Merge the node dependency ya...
@andylizf seems like i cannot assign you to this issue. I assigned myself first How about this format? ```yaml name: pipeline depends_on: train: [] eval: [train] analyze: [eval] visualize: [analyze, train] --- name: train resources: accelerators: V100:8 any_of: - use_spot: true - use_spot: f...
[ { "body": "<!-- Describe the bug report / feature request here -->\r\n\r\nWe should think about the API to specify an DAG of tasks.\r\n\r\nSeveral choices:\r\n\r\n1. Add an attribute to `sky.Task` for predecessor/successor\r\n2. Add a dedicated yaml file for node dependencies\r\n3. Merge the node dependency yam...
340f38404fe5d3ebe35ea430a67cb3377241d1f3
{ "head_commit": "ff528a5de1518737bc2c42ba9ff3f0d61fe41270", "head_commit_message": "add comments for add_edge", "patch_to_review": "diff --git a/examples/dag/diamond.yml b/examples/dag/diamond.yml\nnew file mode 100644\nindex 00000000000..c6791138658\n--- /dev/null\n+++ b/examples/dag/diamond.yml\n@@ -0,0 +1,77 ...
[ { "diff_hunk": "", "line": null, "original_line": null, "original_start_line": null, "path": "examples/dag/diamond.yml", "start_line": null, "text": "@user1:\nI realized this would change the direction of the edge, e.g. for `dependencies` we need to specify its predecessor but not its su...
8992f23822af058e90c9b11eb5f19502f25bacdf
diff --git a/.github/workflows/format.yml b/.github/workflows/format.yml index a19bdcd020d..61f088a4a14 100644 --- a/.github/workflows/format.yml +++ b/.github/workflows/format.yml @@ -7,10 +7,12 @@ on: branches: - master - 'releases/**' + - advanced-dag pull_request: branches: - ...
{ "difficulty": "high", "estimated_review_effort": 4, "problem_domain": "New Feature Additions" }
skypilot-org__skypilot-3730@8a750c1
skypilot-org/skypilot
Python
3,730
[Storage] Make s3 bucket creation log visible to the users
<!-- Describe the changes in this PR --> This resolves #3729 with some refactoring. Tested (run the relevant ones): - [x] Code formatting: `bash format.sh` - [x] Smoke tests `pytest tests/test_smoke.py::TestStorageWithCredentials --aws` - [x] manual `sky launch` with the task yaml mentioned in #3729: ``` $ s...
2024-07-06T22:05:45Z
[Storage] creating s3 bucket fails to display log to user <!-- Describe the bug report / feature request here --> When we create sky managed storages as part of `file_mounts` config from task yaml, we display the created bucket to the users but we currently fail to display for s3 buckets. To reproduce: `test.yaml:`...
[ { "body": "<!-- Describe the bug report / feature request here -->\r\nWhen we create sky managed storages as part of `file_mounts` config from task yaml, we display the created bucket to the users but we currently fail to display for s3 buckets. To reproduce:\r\n\r\n`test.yaml:`\r\n```\r\nfile_mounts:\r\n /s3-...
994d35ae3bf0962e1d21e4db006c2ba1dddcaa0f
{ "head_commit": "8a750c1e764f3c5d25550237695917aaad49ed18", "head_commit_message": "nit", "patch_to_review": "diff --git a/sky/data/storage.py b/sky/data/storage.py\nindex e43406c3951..bb89ad9605e 100644\n--- a/sky/data/storage.py\n+++ b/sky/data/storage.py\n@@ -1368,24 +1368,22 @@ def _create_s3_bucket(self,\n ...
[ { "diff_hunk": "@@ -1368,24 +1368,22 @@ def _create_s3_bucket(self,\n \"\"\"\n s3_client = self.client\n try:\n- if region is None:\n- s3_client.create_bucket(Bucket=bucket_name)\n- else:\n- if region == 'us-east-1':\n- ...
1df84e19304994e8c8fb11bd489b628d7d48f374
diff --git a/sky/data/storage.py b/sky/data/storage.py index e43406c3951..f909df45dd5 100644 --- a/sky/data/storage.py +++ b/sky/data/storage.py @@ -1368,24 +1368,22 @@ def _create_s3_bucket(self, """ s3_client = self.client try: - if region is None: - s3_client.crea...
{ "difficulty": "low", "estimated_review_effort": 2, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-3758@641f8ff
skypilot-org/skypilot
Python
3,758
k8s logtail
<!-- Describe the changes in this PR --> Closes #3429 This swaps the container sleep with a logtailing process so that pod logs now show task logs. These are retrievable via `kubectl logs <POD_NAME>` and also written to `/var/log/pods/*/*/*.log` on the node disk, so it should integrate with most k8s logging solut...
2024-07-17T20:50:54Z
[k8s] Log SkyPilot output to container stdout SkyPilot pods are run with `sleep infinity` to let SkyPilot runtime manage the lifecycle of the pod. As a result, container logs are empty `kubectl get logs`. This causes problems when integrating with other log management systems running on the cluster, which directly s...
[This commit](https://github.com/skypilot-org/skypilot/commit/b95c66491834272b01d68efc955ad4fd9171e4a8) has an alternative solution which routes the stdout to a shared global log file which is tail'd in the container run command. While this also works, it adds the overhead of doubling log size. Perhaps we could hav...
[ { "body": "SkyPilot pods are run with `sleep infinity` to let SkyPilot runtime manage the lifecycle of the pod. As a result, container logs are empty `kubectl get logs`.\r\n\r\nThis causes problems when integrating with other log management systems running on the cluster, which directly scrape from the containe...
a378a08d17a23ef2c45eb16c9759a3635f3e1ff4
{ "head_commit": "641f8ff944c7d46aa52478c721e0a84ab49a3430", "head_commit_message": "lint", "patch_to_review": "diff --git a/sky/templates/kubernetes-ray.yml.j2 b/sky/templates/kubernetes-ray.yml.j2\nindex bd4bafd43d5..8ffb8da8ddb 100644\n--- a/sky/templates/kubernetes-ray.yml.j2\n+++ b/sky/templates/kubernetes-r...
[ { "diff_hunk": "@@ -316,8 +316,48 @@ available_node_types:\n image: {{image_id}}\n # Do not change this command - it keeps the pod alive until it is\n # explicitly killed.\n- command: [\"/bin/bash\", \"-c\", \"--\"]\n- args: ['trap : TERM INT; sleep infinity & wai...
5893896e85e0de98bd32a02cf17fa089e7a01966
diff --git a/sky/templates/kubernetes-ray.yml.j2 b/sky/templates/kubernetes-ray.yml.j2 index bd4bafd43d5..a68c01bc696 100644 --- a/sky/templates/kubernetes-ray.yml.j2 +++ b/sky/templates/kubernetes-ray.yml.j2 @@ -317,7 +317,54 @@ available_node_types: # Do not change this command - it keeps the pod alive unt...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "New Feature Additions" }
skypilot-org__skypilot-3674@879df6a
skypilot-org/skypilot
Python
3,674
[k8s] suppress connection error warnings when disconnected from k8s
<!-- Describe the changes in this PR --> suppresses `urllib3` warnings for `sky show-gpus` if k8s cluster gets disconnected. Closes #3591 before changes ``` (base) gcpuser@k3s-ebd1-head-82q3gnw3-compute:~/skypilot$ sky show-gpus --cloud kubernetes WARNING:urllib3.connectionpool:Retrying (Retry(total=2, conne...
2024-06-18T20:29:35Z
[k8s] Suppress warnings when kubernetes cluster is not reachable We should consider suppressing warnings when the Kubernetes cluster is no longer reachable (e.g., after GKE cluster has been deleted or docker runtime is not running for `sky local up`). Currently, we get warnings like: ``` sky show-gpus --cloud kuberne...
[ { "body": "We should consider suppressing warnings when the Kubernetes cluster is no longer reachable (e.g., after GKE cluster has been deleted or docker runtime is not running for `sky local up`). Currently, we get warnings like:\r\n```\r\nsky show-gpus --cloud kubernetes\r\nWARNING:urllib3.connectionpool:Retr...
a9f5db71e0d5a8ecdf4f7227370829f748abee18
{ "head_commit": "879df6a1b85e799e450381cff7dd76607738aac5", "head_commit_message": "suppress connection error warnings when disconnected from k8s\n\nformat\n\nset urllib3 log level\n\nset to ERROR level", "patch_to_review": "diff --git a/sky/provision/kubernetes/utils.py b/sky/provision/kubernetes/utils.py\ninde...
[ { "diff_hunk": "@@ -16,6 +17,7 @@\n from sky import skypilot_config\n from sky.adaptors import kubernetes\n from sky.provision.kubernetes import network_utils\n+from sky.provision.logging import set_logging_level", "line": null, "original_line": 20, "original_start_line": null, "path": "sky/prov...
4d2550955108cd110c723d3e4f27b9d3d3f96971
diff --git a/sky/adaptors/kubernetes.py b/sky/adaptors/kubernetes.py index 7cdb3ff3059..7f52a099f56 100644 --- a/sky/adaptors/kubernetes.py +++ b/sky/adaptors/kubernetes.py @@ -2,9 +2,11 @@ # pylint: disable=import-outside-toplevel +import logging import os from sky.adaptors import common +from sky.sky_logging...
{ "difficulty": "low", "estimated_review_effort": 2, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-3671@b8001f7
skypilot-org/skypilot
Python
3,671
[Storage] Add storage translation for newly created buckets
Closes #3666. ~We skip the storage name validation because it is already done when the job is submitted from the client and the guardrails are not required for the controller.~ We now translate storage objects which create new buckets to become externally sourced buckets. E.g., ``` /output/: name: my-new...
2024-06-15T19:14:21Z
[Storage] Creating new buckets in `sky jobs launch` fails `sky jobs launch --cloud aws task.yaml` ``` file_mounts: /outputs: name: romil-sky-outputs1 # This is a new unique bucket store: s3 mode: MOUNT run: | echo "Hello, world!" > /outputs/hello.txt ``` Fails with: ``` $ sky jobs lo...
Should also update smoke tests to test this case.
[ { "body": "`sky jobs launch --cloud aws task.yaml`\r\n\r\n```\r\nfile_mounts:\r\n /outputs:\r\n name: romil-sky-outputs1 # This is a new unique bucket\r\n store: s3\r\n mode: MOUNT\r\n\r\nrun: |\r\n echo \"Hello, world!\" > /outputs/hello.txt\r\n```\r\n\r\nFails with:\r\n\r\n```\r\n$ sky jobs logs --...
4821f70b3f4998821dd68c2afcdc7ff61b54ec46
{ "head_commit": "b8001f751ecf9d4b6e94060ec8f5186bd8ae25a1", "head_commit_message": "fix smoke tests", "patch_to_review": "diff --git a/examples/managed_job_with_storage.yaml b/examples/managed_job_with_storage.yaml\nindex ecefccd8b3d..f26b409a1cf 100644\n--- a/examples/managed_job_with_storage.yaml\n+++ b/exampl...
[ { "diff_hunk": "@@ -325,6 +326,13 @@ def __deepcopy__(self, memo):\n \n def _validate_existing_bucket(self):\n \"\"\"Validates the storage fields for existing buckets.\"\"\"\n+ if controller_utils.is_running_on_jobs_controller():\n+ # If running on jobs controller, we skip the chec...
ad6f3c3c5703e0ac0caac048c5cce232802cbc61
diff --git a/examples/managed_job_with_storage.yaml b/examples/managed_job_with_storage.yaml index ecefccd8b3d..61244c16ba0 100644 --- a/examples/managed_job_with_storage.yaml +++ b/examples/managed_job_with_storage.yaml @@ -15,11 +15,17 @@ workdir: ./examples file_mounts: ~/bucket_workdir: - # Change this to ...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-3647@aa494a1
skypilot-org/skypilot
Python
3,647
[UX] Remove warning for gcp check
<!-- Describe the changes in this PR --> Fixes #3060 This also adds the detailed error for GCP operation in the exception raised to make it easier to debug. <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the...
2024-06-06T23:40:12Z
[UX] `sky check` shows warning from GCP package The following message shows during the `sky check`, which may related to some deprecation/update from google API. We should check the cause and fix them. ```console $ sky check Checking credentials to enable clouds for SkyPilot. Checking GCP...INFO:googleapiclient...
Related UX: now that we have many clouds being checked here, most are likely to show as disabled. It'd be great to print a `<checkbox> Enabled clouds: <clouds list>` or something like that at the end of the message.
[ { "body": "The following message shows during the `sky check`, which may related to some deprecation/update from google API. We should check the cause and fix them.\r\n\r\n```console\r\n$ sky check\r\nChecking credentials to enable clouds for SkyPilot.\r\n Checking GCP...INFO:googleapiclient.discovery_cache:fi...
26d902d7e47900bb6b6c897f6fda79047b35df35
{ "head_commit": "aa494a1284d000a6147c031b7bd5394797468091", "head_commit_message": "Add detailed reason in operation failure", "patch_to_review": "diff --git a/sky/clouds/gcp.py b/sky/clouds/gcp.py\nindex 93260533f27..9b2a1494030 100644\n--- a/sky/clouds/gcp.py\n+++ b/sky/clouds/gcp.py\n@@ -742,7 +742,8 @@ def c...
[ { "diff_hunk": "@@ -100,19 +100,20 @@ def _generate_node_name(cluster_name: str, node_suffix: str,\n return node_name\n \n \n-def _log_errors(errors: List[Dict[str, str]], e: Any,\n- zone: Optional[str]) -> None:\n+def _format_message_from_errors(errors: List[Dict[str, str]], e: Any,\n+ ...
9cc675e65267d7e583cff5a0624e893baf4bb595
diff --git a/sky/adaptors/gcp.py b/sky/adaptors/gcp.py index 6465709d42c..9f63bec87ee 100644 --- a/sky/adaptors/gcp.py +++ b/sky/adaptors/gcp.py @@ -21,8 +21,9 @@ def build(service_name: str, version: str, *args, **kwargs): service_name: GCP service name (e.g., 'compute', 'storagetransfer'). version: ...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Dependency Updates & Env Compatibility" }
skypilot-org__skypilot-3575@994c4fc
skypilot-org/skypilot
Python
3,575
[Core] Install SkyPilot runtime in separate env
<!-- Describe the changes in this PR --> ~~Blocked by #3572~~ Moved #2801 here due to the merging issues and our latest support for explicitly python / ray path. **This PR enables:** 1. Users to **install dependencies in the base environment** without worrying about conflicting with SkyPilot runtime. 2. User...
2024-05-21T18:28:42Z
[core] Ray version change breaks SkyPilot cluster If the user's setup installs a newer version of ray than the one running on SkyPilot remote cluster, SkyPilot will get stuck at streaming logs. Many packages (e.g., `vllm==0.2.0`) tend to install a newer version of ray. SkyPilot should: 1. Protect the ray installat...
We should install skypilot's remote dependency in the environment other than base to avoid this kind of issue. Another user encountered similar issue due to installing some package in the base environment. > We should install skypilot's remote dependency in the environment other than base to avoid this kind of issue. A...
[ { "body": "If the user's setup installs a newer version of ray than the one running on SkyPilot remote cluster, SkyPilot will get stuck at streaming logs. Many packages (e.g., `vllm==0.2.0`) tend to install a newer version of ray. \r\n\r\nSkyPilot should:\r\n1. Protect the ray installation/use a custom env for ...
ccfc3b1e55cc88d8527e4270706e1dd6d4d595d3
{ "head_commit": "994c4fca511aa56b3f3fa9b2b8228f33c39f2e05", "head_commit_message": "fix event", "patch_to_review": "diff --git a/llm/axolotl/axolotl-docker.yaml b/llm/axolotl/axolotl-docker.yaml\nnew file mode 100644\nindex 00000000000..9cec1d1f331\n--- /dev/null\n+++ b/llm/axolotl/axolotl-docker.yaml\n@@ -0,0 +...
[ { "diff_hunk": "@@ -46,3 +35,4 @@ envs:\n \n \n \n+4", "line": null, "original_line": 38, "original_start_line": null, "path": "llm/axolotl/axolotl-spot.yaml", "start_line": null, "text": "@user1:\naccidentally?\n\n@author:\nDone. Thanks!" } ]
71ba100321810fc4d10cfcc2a847405c3f86f52e
diff --git a/llm/axolotl/axolotl-docker.yaml b/llm/axolotl/axolotl-docker.yaml new file mode 100644 index 00000000000..b883ebdde46 --- /dev/null +++ b/llm/axolotl/axolotl-docker.yaml @@ -0,0 +1,29 @@ +# Usage: +# HF_TOKEN=abc sky launch -c axolotl axolotl.yaml --env HF_TOKEN -y -i30 --down + +name: axolotl + +resourc...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-3567@a4aa491
skypilot-org/skypilot
Python
3,567
[Core] Allow nonexistent cloud in candidate resources and speed up optimization
<!-- Describe the changes in this PR --> Closes #3565 ## Allow nonexistent cloud ### With at least one cloud enabled Without kubernetes and IBM enabled: ```yaml resources: accelerators: A100:8 ordered: - cloud: kubernetes - cloud: aws use_spot: true - cloud: ibm ``` Master: <d...
2024-05-20T07:17:14Z
Certain inputs to `sky launch` would trigger `sky check`, making it slow <!-- Describe the bug report / feature request here --> ``` resources: accelerators: A100:8 ordered: - cloud: kubernetes - cloud: aws use_spot: true ``` My `sky check` has AWS, GCP, ... enabled, no k8s. On latest m...
[ { "body": "<!-- Describe the bug report / feature request here -->\r\n\r\n```\r\nresources:\r\n accelerators: A100:8\r\n ordered:\r\n - cloud: kubernetes\r\n - cloud: aws\r\n use_spot: true\r\n```\r\nMy `sky check` has AWS, GCP, ... enabled, no k8s.\r\n\r\nOn latest main branch: 19e8ed16a, it took...
6a624b2fa3533496bbed3bf013ae608e834b807f
{ "head_commit": "a4aa49125c85f42e83b4092b53da4974f0056cb1", "head_commit_message": "revert is_same_cloud", "patch_to_review": "diff --git a/sky/check.py b/sky/check.py\nindex f4ecd5a8b18..e8a61317d63 100644\n--- a/sky/check.py\n+++ b/sky/check.py\n@@ -18,7 +18,7 @@\n def check(\n quiet: bool = False,\n v...
[ { "diff_hunk": "@@ -992,18 +994,18 @@ def ordinal_number(n):\n for task_id in range(len(dag.tasks)):\n task = dag.tasks[task_id]\n if isinstance(task.resources, list):\n+ # For ordered resources, we try the resources in the order\n+ # specified by th...
97572c55657ce357067ac3ea247309cae0e6ffd5
diff --git a/sky/check.py b/sky/check.py index f4ecd5a8b18..e8a61317d63 100644 --- a/sky/check.py +++ b/sky/check.py @@ -18,7 +18,7 @@ def check( quiet: bool = False, verbose: bool = False, - clouds: Optional[Tuple[str]] = None, + clouds: Optional[Iterable[str]] = None, ) -> None: echo = (lambda ...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "Performance Optimizations" }
skypilot-org__skypilot-4564@c85588b
skypilot-org/skypilot
Python
4,564
Introduce High Availability Service Controller
<!-- Describe the changes in this PR --> Resolves #4563 <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the relevant ones): - [x] Code formatting: `bash format.sh` - [x] Any manual or new tests for this PR ...
2025-01-15T15:01:02Z
[Serve] Make SkyServe Controller More Persistent # Design Docs: Persistent Service in SkyPilot ## Overview The Persistent Service feature aims to enhance SkyServe's reliability by enabling service recovery after controller failures or system shutdowns. This design addresses how to restore services with their previo...
[ { "body": "# Design Docs: Persistent Service in SkyPilot\r\n\r\n## Overview\r\nThe Persistent Service feature aims to enhance SkyServe's reliability by enabling service recovery after controller failures or system shutdowns. This design addresses how to restore services with their previous states preserved, mak...
d3e3be380ebcb90d7d3b94ccdede47b3b7469bff
{ "head_commit": "c85588b99166b9c672903e7a750552037d176280", "head_commit_message": "refactor: common util judging if controller by cluster name", "patch_to_review": "diff --git a/docs/source/reference/config.rst b/docs/source/reference/config.rst\nindex a76dc473206..a18ea7ef380 100644\n--- a/docs/source/referenc...
[ { "diff_hunk": "@@ -605,6 +607,15 @@ def check_features_are_supported(\n unsupported_features2reason = cls._unsupported_features_for_resources(\n resources)\n \n+ # pylint: disable=import-outside-toplevel\n+ from sky.clouds import kubernetes\n+ if cls is not kubernetes.K...
47ce8567bd0ae5fcb8e26162ebd5b2ac678a7a98
diff --git a/docs/source/reference/config.rst b/docs/source/reference/config.rst index cc64bbea66b..70c0d3b0b2b 100644 --- a/docs/source/reference/config.rst +++ b/docs/source/reference/config.rst @@ -14,6 +14,8 @@ This configuration applies to all new clusters and does not affect existing clus Syntax ------ +For d...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "New Feature Additions" }
skypilot-org__skypilot-3426@2b717ba
skypilot-org/skypilot
Python
3,426
[UX] Add cluster info in task envs
<!-- Describe the changes in this PR --> Fixes #3413 We are adding a new env var for the user to get the cluster name, cloud, region, and zone. ``` 'SKYPILOT_CLUSTER_INFO': json.dumps({ 'cluster_name': handle.cluster_name, 'cloud': handle.launched_resources.cloud, ...
2024-04-05T05:35:55Z
[Observability] Expose new env vars for: cloud, region, cluster name <!-- Describe the bug report / feature request here --> User asked: > Can skypilot expose more env variables to be available besides SKYPILOT_TASK_ID? Something additional that could be useful: Name of the cluster Name of the cloud Region withi...
[ { "body": "<!-- Describe the bug report / feature request here -->\r\n\r\nUser asked:\r\n> Can skypilot expose more env variables to be available besides SKYPILOT_TASK_ID? Something additional that could be useful:\r\nName of the cluster\r\nName of the cloud\r\nRegion within the cloud\r\nThese would help us bet...
48b8ca978e936412db757800b58450c19ae49976
{ "head_commit": "2b717bab326632e6a12b026bf8070347d6046170", "head_commit_message": "format", "patch_to_review": "diff --git a/docs/source/running-jobs/environment-variables.rst b/docs/source/running-jobs/environment-variables.rst\nindex 8bc5e667e05..d3ef4c2e531 100644\n--- a/docs/source/running-jobs/environment-...
[ { "diff_hunk": "@@ -118,6 +118,9 @@ Environment variables for ``setup``\n - sky-2023-07-06-21-18-31-563597_myclus_1\n \n For managed spot jobs: sky-managed-2023-07-06-21-18-31-563597_my-job-name_1-0\n+ * - ``SKYPILOT_CLUSTER_INFO``\n+ - A JSON string containing information about the clust...
955cc749e289462f9337b362ac3b92883bb5729f
diff --git a/docs/source/running-jobs/environment-variables.rst b/docs/source/running-jobs/environment-variables.rst index 8bc5e667e05..16502f70818 100644 --- a/docs/source/running-jobs/environment-variables.rst +++ b/docs/source/running-jobs/environment-variables.rst @@ -118,6 +118,9 @@ Environment variables for ``set...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "New Feature Additions" }
skypilot-org__skypilot-3538@930f148
skypilot-org/skypilot
Python
3,538
[Core] Add `sky jobs logs --name` and fix sky job logs spinner
<!-- Describe the changes in this PR --> This adds support for `sky jobs logs --name`, fixes #3599, and add test for `sky job logs` This is to prevent issue like #3531 with the smoke test. <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the...
2024-05-10T17:13:43Z
[Jobs] `sky jobs logs` failed to show status spinner when the job is starting <!-- Describe the bug report / feature request here --> Our recent master fails to show the status spinner in `sky jobs logs` when the job is starting. We should bisect to find the reason.
[ { "body": "<!-- Describe the bug report / feature request here -->\r\n\r\nOur recent master fails to show the status spinner in `sky jobs logs` when the job is starting. We should bisect to find the reason.\r\n\r\n\r\n", "number": 3599, "title": "[Jobs] `sky jobs logs` failed to show status spinner when...
e006a791162071ccda9661a0cfc7250d87fb1e12
{ "head_commit": "930f14816d47ea107a444bc5fc7db83e44c40abc", "head_commit_message": "format", "patch_to_review": "diff --git a/examples/managed_job.yaml b/examples/managed_job.yaml\nindex 4bfcb63f40a..30ad2db287a 100644\n--- a/examples/managed_job.yaml\n+++ b/examples/managed_job.yaml\n@@ -6,6 +6,7 @@ setup: |\n ...
[ { "diff_hunk": "@@ -448,18 +449,48 @@ def stream_logs_by_id(job_id: int, follow: bool = True) -> str:\n return ''\n \n \n-def stream_logs_by_name(job_name: str, follow: bool = True) -> str:\n- \"\"\"Stream logs by name.\"\"\"\n- job_ids = managed_job_state.get_nonterminal_job_ids_by_name(job_name)\n- ...
5e8b30e2337b2ab4a05fa82c3673b022f75dace6
diff --git a/examples/managed_job.yaml b/examples/managed_job.yaml index 4bfcb63f40a..30ad2db287a 100644 --- a/examples/managed_job.yaml +++ b/examples/managed_job.yaml @@ -6,6 +6,7 @@ setup: | run: | conda env list + echo "start counting" python -u - << EOF import time import tqdm diff --git a/sky/back...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-3424@3a52595
skypilot-org/skypilot
Python
3,424
[Spot] Add spot job name in the `SKYPILOT_TASK_ID` env var
<!-- Describe the changes in this PR --> Fixes #3391 This PR adds the task name in the `SKYPILOT_TASK_ID`. <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the relevant ones): - [ ] Code formatting: `bash f...
2024-04-05T04:23:14Z
[Spot/UX] Make spot job name part of `SKYPILOT_TASK_ID` ``` name: my-task-name resources: cpus: 2+ use_spot: true run: | echo SKYPILOT_TASK_ID: $SKYPILOT_TASK_ID ``` `sky launch -c my-cluster t-name.yaml --down` - `SKYPILOT_TASK_ID: sky-2024-03-29-08-33-56-952900_my-cluster_id-1` -- note the cluste...
[ { "body": "```\r\nname: my-task-name\r\n\r\nresources:\r\n cpus: 2+\r\n use_spot: true\r\n\r\nrun: |\r\n echo SKYPILOT_TASK_ID: $SKYPILOT_TASK_ID\r\n```\r\n\r\n`sky launch -c my-cluster t-name.yaml --down`\r\n- `SKYPILOT_TASK_ID: sky-2024-03-29-08-33-56-952900_my-cluster_id-1` -- note the cluster (rather tha...
7672360fc101d8d06469141c51b1439653299fcc
{ "head_commit": "3a52595cb58eb7e34189758c96069c431bb6e93d", "head_commit_message": "format", "patch_to_review": "diff --git a/docs/source/examples/spot-jobs.rst b/docs/source/examples/spot-jobs.rst\nindex 6aa5606968e..22ddcee58d5 100644\n--- a/docs/source/examples/spot-jobs.rst\n+++ b/docs/source/examples/spot-j...
[ { "diff_hunk": "@@ -369,6 +369,12 @@ The above YAML file defines a pipeline with two tasks. The first :code:`name: pi\n \n To submit the pipeline, the same command :code:`sky spot launch` is used. The pipeline will be automatically launched and monitored by SkyPilot. You can check the status of the pipeline wit...
45c42689227e560dfc5d02d5c7e7214487ce166f
diff --git a/docs/source/examples/spot-jobs.rst b/docs/source/examples/spot-jobs.rst index 6aa5606968e..5940e404bb3 100644 --- a/docs/source/examples/spot-jobs.rst +++ b/docs/source/examples/spot-jobs.rst @@ -44,7 +44,7 @@ We can launch it with the following: .. code-block:: yaml # bert_qa.yaml - name: bert_qa +...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "New Feature Additions" }
skypilot-org__skypilot-3362@ccf99c2
skypilot-org/skypilot
Python
3,362
[Core] Support custom image on Azure
<!-- Describe the changes in this PR --> This PR adds support for custom images on Azure. Partially fixes #2910. Related: skypilot-org/skypilot-catalog#66 <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the r...
2024-03-24T03:15:46Z
Cannot choose Custom V-net and custom images in Azure I cannot choose Custom V-net and custom images in Azure, is there a specific format I have to use, or is the feature not available for Azure? (sky) skypilot@kenadyinampudi:$ sky launch skypilot_azure.yaml ValueError: Invalid config YAML (/home/skypilot/.sky/co...
> [ ] `image_id` support for Azure, like what we have for AWS/GCP. Another user bumped this.
[ { "body": "I cannot choose Custom V-net and custom images in Azure, is there a specific format I have to use, or is the feature not available for Azure?\r\n\r\n\r\n(sky) skypilot@kenadyinampudi:$ sky launch skypilot_azure.yaml\r\nValueError: Invalid config YAML (/home/skypilot/.sky/config.yaml): The following f...
82c50f5c281fc88eb417dfd928dbe570da99aaf4
{ "head_commit": "ccf99c24d7d5a3efbc3a409b04e879bdf71ad4c7", "head_commit_message": "retry tests after merging catalog", "patch_to_review": "diff --git a/docs/source/reference/yaml-spec.rst b/docs/source/reference/yaml-spec.rst\nindex 402dfef4cfe..782fd9fd208 100644\n--- a/docs/source/reference/yaml-spec.rst\n+++...
[ { "diff_hunk": "@@ -137,6 +137,51 @@ def get_egress_cost(self, num_gigabytes: float):\n def is_same_cloud(self, other):\n return isinstance(other, Azure)\n \n+ @classmethod\n+ def get_image_size(cls, image_id: str, region: Optional[str]) -> float:\n+ if region is None:\n+ # T...
cc24daa59ae9290ca386a19f1f1636db9d0c9ff6
diff --git a/docs/source/reference/yaml-spec.rst b/docs/source/reference/yaml-spec.rst index 402dfef4cfe..782fd9fd208 100644 --- a/docs/source/reference/yaml-spec.rst +++ b/docs/source/reference/yaml-spec.rst @@ -194,6 +194,10 @@ Available fields: # Or machine image: https://cloud.google.com/compute/docs/machine...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-3326@eaa2ac2
skypilot-org/skypilot
Python
3,326
[Core] Use original Python path for SkyPilot runtime
<!-- Describe the changes in this PR --> Fixes #3324 master ([823999a](https://github.com/skypilot-org/skypilot/commit/823999af850ee93138f45d01abba6c54a93d3c1e)) `for i in {1..5}; do time sky launch -c test-gcp-$i --cloud gcp --cpus 2 -y; done` ``` 1m59.195s 2m0.108s 1m56.507s 1m54.553s 1m54.242s ``` me...
2024-03-18T09:15:20Z
[Core] User changing default env will make job submission fail To reproduce: ```yaml resources: cpus: 2+ setup: | conda activate myenv if [ $? -ne 0 ]; then conda create -n myenv python=3.10 -y conda activate myenv fi grep -qxF 'conda activate myenv' ~/.bashrc || echo "conda activate mye...
[ { "body": "To reproduce:\r\n```yaml\r\nresources:\r\n cpus: 2+\r\n\r\nsetup: |\r\n conda activate myenv\r\n if [ $? -ne 0 ]; then\r\n conda create -n myenv python=3.10 -y\r\n conda activate myenv\r\n fi\r\n\r\n grep -qxF 'conda activate myenv' ~/.bashrc || echo \"conda activate myenv\" >> ~/.bashrc\r...
823999af850ee93138f45d01abba6c54a93d3c1e
{ "head_commit": "eaa2ac22afb6bf8fe470955ba4462077aec8fb8a", "head_commit_message": "Fix test", "patch_to_review": "diff --git a/sky/backends/backend_utils.py b/sky/backends/backend_utils.py\nindex dd7711fe326..9099e1b999b 100644\n--- a/sky/backends/backend_utils.py\n+++ b/sky/backends/backend_utils.py\n@@ -850,7...
[ { "diff_hunk": "@@ -903,6 +903,8 @@ def write_cluster_config(\n 'ray_temp_dir': constants.SKY_REMOTE_RAY_TEMPDIR,\n 'dump_port_command': dump_port_command,\n # Ray version.\n+ 'RAY_CMD': constants.SKY_RAY_CMD,\n+ 'SKY_PIP_CMD': consta...
4874a77a38ebd36d88fa312ae4b05a0c8409c267
diff --git a/sky/backends/backend_utils.py b/sky/backends/backend_utils.py index dd7711fe326..850dc2009ca 100644 --- a/sky/backends/backend_utils.py +++ b/sky/backends/backend_utils.py @@ -850,7 +850,7 @@ def write_cluster_config( # Dump the Ray ports to a file for Ray job submission dump_port_command = ( -...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-3319@d09a6a1
skypilot-org/skypilot
Python
3,319
[Storage] Use task's region for initializing new stores
Closes #3114. Uses the optimizer's `best_resources` or the the tasks' resource spec to infer the region for the bucket, if available. Note that currently the bucket is eagerly provisioned in the first region that will be tried by the provisioner. If failover is trigerred, the bucket and compute may lie in a differen...
2024-03-16T10:07:27Z
[Bug/Feature Request] sky serve doesn't respect regions bucket It seems sky serve doesn't respect the settings for the region. sky serve created a bucket in us-east-2 and tried to create a controller in the same regions. UPDATE: controller can be configured in config.yaml via serve section. (Luckily it failed for me...
@manzke Thanks for reporting this! Actually the `resources` section under the Service YAML is the resources requirement for service replicas, not the controller. To specify controller resources, pls check here: https://skypilot.readthedocs.io/en/latest/serving/sky-serve.html#customizing-skyserve-controller-resources Yo...
[ { "body": "It seems sky serve doesn't respect the settings for the region. sky serve created a bucket in us-east-2 and tried to create a controller in the same regions. UPDATE: controller can be configured in config.yaml via serve section.\r\n\r\n(Luckily it failed for me, because the specified vpc is only avai...
9d6bf82224a473b5f5357a91c9e62d07e71a8e01
{ "head_commit": "d09a6a13f90d13a6410b31876e36fdff0775facb", "head_commit_message": "make stores region aware", "patch_to_review": "diff --git a/sky/data/storage.py b/sky/data/storage.py\nindex 9356a50f365..081d3b0425a 100644\n--- a/sky/data/storage.py\n+++ b/sky/data/storage.py\n@@ -791,7 +791,9 @@ def from_meta...
[ { "diff_hunk": "@@ -857,15 +857,16 @@ def update_storage_mounts(\n task_storage_mounts.update(storage_mounts)\n return self.set_storage_mounts(task_storage_mounts)\n \n- def get_preferred_store_type(self) -> storage_lib.StoreType:\n+ def get_preferred_store(self) -> Tuple[storage_lib.Store...
ba971add386af1202931083fe414feb33091b8ca
diff --git a/sky/data/data_utils.py b/sky/data/data_utils.py index 034e6c78180..21717ec739a 100644 --- a/sky/data/data_utils.py +++ b/sky/data/data_utils.py @@ -90,12 +90,15 @@ def split_cos_path(s3_path: str) -> Tuple[str, str, str]: return bucket_name, data_path, region -def create_s3_client(region: str = 'u...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-3363@7ab4b7c
skypilot-org/skypilot
Python
3,363
[Core] Support choosing cloud for Spot controller
<!-- Describe the changes in this PR --> Addresses the new comment under #3231. <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the relevant ones): - [x] Code formatting: `bash format.sh` - [x] Any manual or...
2024-03-25T07:20:40Z
[Serve] Fail to launch controller when no resources is specified for a service when firstly used <!-- Describe the bug report / feature request here --> This issue affects all first-time SkyServe user who don't have serve controller started before To reproduce: 1. with no serve controller 2. start the following...
[ { "body": "<!-- Describe the bug report / feature request here -->\r\n\r\nThis issue affects all first-time SkyServe user who don't have serve controller started before\r\n\r\nTo reproduce:\r\n1. with no serve controller\r\n2. start the following service\r\n\r\n```yaml\r\nresources:\r\n accelerators: L4\r\n p...
19df9d3c52762a3c67b2f5491ffc173b6801858c
{ "head_commit": "7ab4b7c58be3de7aec3f401fd5a7e4f773d7a885", "head_commit_message": "fix unittest", "patch_to_review": "diff --git a/sky/check.py b/sky/check.py\nindex 7e129e4fde5..70d561fba15 100644\n--- a/sky/check.py\n+++ b/sky/check.py\n@@ -124,7 +124,7 @@ def get_cloud_credential_file_mounts(\n file_moun...
[ { "diff_hunk": "@@ -150,11 +137,12 @@ def up(\n serve_utils.generate_remote_config_yaml_file_name(service_name))\n controller_log_file = (\n serve_utils.generate_remote_controller_log_file_name(service_name))\n- controller_resources_in_config = (\n- controller_u...
9cddf933e23849d9602bf577ad14e88e08b07a2f
diff --git a/sky/check.py b/sky/check.py index 9c28c43ec53..6818d80f3bf 100644 --- a/sky/check.py +++ b/sky/check.py @@ -130,7 +130,7 @@ def get_cloud_credential_file_mounts( file_mounts = {} for cloud in enabled_clouds: if (excluded_clouds is not None and - clouds.cloud_in_list(cloud,...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-3288@b3ff5d9
skypilot-org/skypilot
Python
3,288
[Serve/Spot] Allow spot queue/cancel/logs during controller INIT state
<!-- Describe the changes in this PR --> Fixes #1592, #3285 TODO: - [x] Handle the case, when the controller is actually abnormal (autostopped or manually terminated) <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Test...
2024-03-08T09:49:42Z
[UX] `sky spot queue/logs` unavailable when there are multiple spot jobs being submitted From our users: > The `sky spot queue/logs` will remain unavailable when multiple spot jobs are being submitted. Two possible ways to solve this: 1. Add an option (or change the meaning of `sky spot queue -r`) so that it waits...
The second option may lead to some issue that the `skypilot` package is being installed on the cluster. This is a recurring request from our users. Elevating the priority.
[ { "body": "From our users:\r\n> The `sky spot queue/logs` will remain unavailable when multiple spot jobs are being submitted.\r\n\r\nTwo possible ways to solve this:\r\n1. Add an option (or change the meaning of `sky spot queue -r`) so that it waits until the controller is not in INIT mode.\r\n2. Try to run th...
a7fdfd240f610f955180ea844c545ca68569fe61
{ "head_commit": "b3ff5d9416701cdc299438157234d94f130532fa", "head_commit_message": "address comments", "patch_to_review": "diff --git a/.github/workflows/pytest.yml b/.github/workflows/pytest.yml\nindex 019b2cb4564..87d8cea9f16 100644\n--- a/.github/workflows/pytest.yml\n+++ b/.github/workflows/pytest.yml\n@@ -2...
[ { "diff_hunk": "@@ -639,7 +640,7 @@ def tail_logs(\n 'target=CONTROLLER/LOAD_BALANCER.')\n handle = backend_utils.is_controller_accessible(\n controller_type=controller_utils.Controllers.SKY_SERVE_CONTROLLER,\n- stopped_message='No in-progress services.')\n+ ...
cac70939908baa5e910dd4cd121d3d6a94a87441
diff --git a/.github/workflows/pytest.yml b/.github/workflows/pytest.yml index 019b2cb4564..87d8cea9f16 100644 --- a/.github/workflows/pytest.yml +++ b/.github/workflows/pytest.yml @@ -27,7 +27,7 @@ jobs: - tests/test_optimizer_random_dag.py - tests/test_storage.py - tests/test_wheels.p...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "New Feature Additions" }
skypilot-org__skypilot-3078@287f2c6
skypilot-org/skypilot
Python
3,078
[Serve] Support env var in service yaml
<!-- Describe the changes in this PR --> Fixes https://github.com/skypilot-org/skypilot/issues/3048 <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the relevant ones): - [x] Code formatting: `bash format.sh` -...
2024-02-04T04:54:00Z
[Serve] Support env var in `service` field When creating an example for serving llm, it will be much easier if the service field can support the following: ```yaml service: readiness_probe: path: /v1/chat/completions post_data: model: $MODEL_NAME messages: - role: user ...
Agree! would be great to have
[ { "body": "When creating an example for serving llm, it will be much easier if the service field can support the following:\r\n\r\n```yaml\r\nservice:\r\n readiness_probe:\r\n path: /v1/chat/completions\r\n post_data:\r\n model: $MODEL_NAME\r\n messages:\r\n - role: user\r\n con...
3ffef36a23b020487b71bf27afc274cbf8411110
{ "head_commit": "287f2c681c465036d5647fadb93c17a9944233e2", "head_commit_message": "[serve] add env vars to smoke test", "patch_to_review": "diff --git a/sky/task.py b/sky/task.py\nindex 36de66370d0..17da10c95b1 100644\n--- a/sky/task.py\n+++ b/sky/task.py\n@@ -72,11 +72,11 @@ def _is_valid_name(name: Optional[s...
[ { "diff_hunk": "@@ -1,7 +1,10 @@\n service:\n- readiness_probe: /v1/models\n+ readiness_probe: /${API_VERSION}/models", "line": null, "original_line": 2, "original_start_line": null, "path": "tests/skyserve/llm/service.yaml", "start_line": null, "text": "@user1:\nIt will be great to us...
46bbb5b2fe62032a5938f7af670055b74955f062
diff --git a/sky/task.py b/sky/task.py index 36de66370d0..17da10c95b1 100644 --- a/sky/task.py +++ b/sky/task.py @@ -72,11 +72,11 @@ def _is_valid_name(name: Optional[str]) -> bool: return bool(re.fullmatch(_VALID_NAME_REGEX, name)) -def _fill_in_env_vars_in_file_mounts( - file_mounts: Dict[str, Any], +def ...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "New Feature Additions" }
skypilot-org__skypilot-3205@0696eba
skypilot-org/skypilot
Python
3,205
[UX] Reset autostop timer whenever autostop config changes on remote machine
<!-- Describe the changes in this PR --> Fixes #1511 <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the relevant ones): - [ ] Code formatting: `bash format.sh` - [ ] Any manual or new tests for this PR (ple...
2024-02-21T19:06:03Z
[Autostop] Need to reset the timer for every execution stage Our user observed that the spot controller is autostopped, during the submission of a spot job (during setting up the controller). It is possible because we only reset the autostop timer, when the controller process is submitted to the ray job. One way to ...
This issue is stale because it has been open 120 days with no activity. Remove stale label or comment or this will be closed in 10 days. This issue was closed because it has been stalled for 10 days with no activity. This issue is stale because it has been open 120 days with no activity. Remove stale label or comment o...
[ { "body": "Our user observed that the spot controller is autostopped, during the submission of a spot job (during setting up the controller). It is possible because we only reset the autostop timer, when the controller process is submitted to the ray job.\r\n\r\nOne way to mitigate it is to reset the autostop t...
40fa245d8206ca911ed138a8e52befae4553e015
{ "head_commit": "0696ebae9988440e14b2b3ff34ef0823367b9eb6", "head_commit_message": "update smoke test", "patch_to_review": "diff --git a/sky/cli.py b/sky/cli.py\nindex 6feb97690fa..e0069a671ab 100644\n--- a/sky/cli.py\n+++ b/sky/cli.py\n@@ -2559,17 +2559,12 @@ def autostop(\n \n - The cluster has restarted.\...
[ { "diff_hunk": "@@ -2559,17 +2559,12 @@ def autostop(\n \n - The cluster has restarted.\n \n- - An autostop is set when there is no active setting. (Namely, either\n- there's never any autostop setting set, or the previous autostop setting\n- was canceled.) This is useful for restarting the aut...
c0d442f97b34e15ae81dd0b44c4f0ef25bfd3aec
diff --git a/sky/cli.py b/sky/cli.py index 6feb97690fa..f91c88cce19 100644 --- a/sky/cli.py +++ b/sky/cli.py @@ -2559,18 +2559,18 @@ def autostop( - The cluster has restarted. - - An autostop is set when there is no active setting. (Namely, either - there's never any autostop setting set, or the previo...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-3132@008b51d
skypilot-org/skypilot
Python
3,132
[Core] Wait `ray status` to be ready
<!-- Describe the changes in this PR --> Fixes #3129 This PR also raises proper error for the case when Azure have an invalid state string. Fixes #3134 <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the rele...
2024-02-09T05:35:38Z
[Spot] Spot jobs on Azure always says it is preempted before submitting the job A user reported that a managed spot job on Azure always says it is preempted after the cluster is launched before submitting the job. This is likely because the `ray cluster` takes longer than other clouds to initalize before `ray status...
To reproduce: `sky launch --cpus 2+ --cloud azure -c test-azure-query; sky status -r test-azure-query`
[ { "body": "A user reported that a managed spot job on Azure always says it is preempted after the cluster is launched before submitting the job.\r\n\r\nThis is likely because the `ray cluster` takes longer than other clouds to initalize before `ray status` can return the correct value of the cluster.", "num...
d2b2118c99392917b54811f5672821824d98470d
{ "head_commit": "008b51dcf3b4c1990b3bb24b83a92a830c131123", "head_commit_message": "add comment", "patch_to_review": "diff --git a/sky/backends/backend_utils.py b/sky/backends/backend_utils.py\nindex 248ddb1dc17..55666ea1d3f 100644\n--- a/sky/backends/backend_utils.py\n+++ b/sky/backends/backend_utils.py\n@@ -88...
[ { "diff_hunk": "@@ -169,9 +169,7 @@ setup_commands:\n # current num items (num SSH connections): 2\n head_start_ray_commands:\n # NOTE: --disable-usage-stats in `ray start` saves 10 seconds of idle wait.\n- - ray stop; RAY_SCHEDULER_EVENTS=0 RAY_DEDUP_LOGS=0 ray start --disable-usage-stats --head --port={{...
9153584d7ab1f481618f653e3b2965aee7b735bf
diff --git a/sky/backends/backend_utils.py b/sky/backends/backend_utils.py index 248ddb1dc17..0711a694325 100644 --- a/sky/backends/backend_utils.py +++ b/sky/backends/backend_utils.py @@ -882,6 +882,10 @@ def write_cluster_config( 'dump_port_command': dump_port_command, # Ray version....
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-3017@1f526ee
skypilot-org/skypilot
Python
3,017
[LLM] Fix mixtral example for Azure
<!-- Describe the changes in this PR --> Fixes #2905 Fix mixtral example for Azure by removing the problemetic nccl config file on Azure. With this PR, after #2434 is merged, we should be able to change the examples to `disk_tier: best`, so that Azure is allowed to be part of the candidates. <!-- Describe the...
2024-01-23T22:17:03Z
[Example] Mixtral example fail to work on Azure VM due to NCCL error `sky launch --disk-tier none -c test-mixtral --cloud azure llm/mixtral/serve.yaml` ``` I 12-27 11:40:38 log_lib.py:431] Start streaming logs for job 1. INFO: Tip: use Ctrl-C to exit log streaming (task will not be killed). INFO: Waiting for tas...
@Michaelvll have you resolved it? Thanks for pinning this @yangyingxiang! Currently we did not figure out the reason for this yet. It seems to be an issue with Azure's GPU setup, and we found people also experiencing this issue: https://github.com/NVIDIA/nccl/issues/1101. We will keep an eye on this issue, and see...
[ { "body": "`sky launch --disk-tier none -c test-mixtral --cloud azure llm/mixtral/serve.yaml` \r\n\r\n```\r\nI 12-27 11:40:38 log_lib.py:431] Start streaming logs for job 1.\r\nINFO: Tip: use Ctrl-C to exit log streaming (task will not be killed).\r\nINFO: Waiting for task resources on 1 node. This will block i...
47206ace8626c464cadeec402a1fadd097d35469
{ "head_commit": "1f526eebec442acbb4585409a9d268db1dd2a83d", "head_commit_message": "Add considered resources", "patch_to_review": "diff --git a/llm/mixtral/README.md b/llm/mixtral/README.md\nindex 68cc97820d5..8bef6e70db4 100644\n--- a/llm/mixtral/README.md\n+++ b/llm/mixtral/README.md\n@@ -19,6 +19,30 @@ resour...
[ { "diff_hunk": "@@ -34,6 +34,9 @@ setup: |\n pip list | grep megablocks || pip install megablocks\n \n run: |\n+ # Remove the default nccl.conf which causes failure on Azure\n+ sudo mv /etc/nccl.conf /etc/nccl.conf.bak || true", "line": null, "original_line": 38, "original_start_line": null, ...
79ea12301156229d4b2558121b29565d84bade90
diff --git a/llm/mixtral/README.md b/llm/mixtral/README.md index 68cc97820d5..8bef6e70db4 100644 --- a/llm/mixtral/README.md +++ b/llm/mixtral/README.md @@ -19,6 +19,30 @@ resources: accelerators: {A100:4, A100:8, A100-80GB:2, A100-80GB:4, A100-80GB:8} ``` +The following is the example output of the optimizer: + ...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Dependency Updates & Env Compatibility" }
skypilot-org__skypilot-2969@72be722
skypilot-org/skypilot
Python
2,969
[Core] Fix Inferentia job scheduling
<!-- Describe the changes in this PR --> Fixes #2968. Bug fixed is when we are submitting a job request, do not request `GPU: <count>` resources. <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the relevant one...
2024-01-10T00:41:57Z
Job submission on Inferentia instances blocks ``` sky launch -t inf2.8xlarge -i60 --down echo hi ``` blocks: ``` .. I 01-09 15:41:36 provisioner.py:464] Successfully provisioned cluster: inf I 01-09 15:41:42 cloud_vm_ray_backend.py:3320] Job submitted with Job ID: 1 I 01-09 23:41:44 log_lib.p...
[ { "body": "```\r\nsky launch -t inf2.8xlarge -i60 --down echo hi \r\n```\r\nblocks:\r\n```\r\n..\r\nI 01-09 15:41:36 provisioner.py:464] Successfully provisioned cluster: inf\r\nI 01-09 15:41:42 cloud_vm_ray_backend.py:3320] Job submitted with Job ID: 1\r\nI 01-09 23:41:44 log_lib.py:431] Star...
6fd5bf298f9d830bb3fbafee0e131ef688142f86
{ "head_commit": "72be722bc270baf15f55a330730a6b944fb1e027", "head_commit_message": "Add smoke", "patch_to_review": "diff --git a/sky/backends/cloud_vm_ray_backend.py b/sky/backends/cloud_vm_ray_backend.py\nindex 1e6085ec222..e40ad1248ac 100644\n--- a/sky/backends/cloud_vm_ray_backend.py\n+++ b/sky/backends/cloud...
[ { "diff_hunk": "@@ -325,8 +325,8 @@ def add_gang_scheduling_placement_group_and_setup(\n gpu_dict = {'GPU': acc_count}\n # gpu_dict should be empty when the accelerator is not GPU.\n # FIXME: This is a hack to make sure that we do not reserve\n- # GPU when requesti...
4e2b182e1bf960e43aabfbd0688919f9756e3ee3
diff --git a/sky/backends/cloud_vm_ray_backend.py b/sky/backends/cloud_vm_ray_backend.py index 1e6085ec222..be1cb0d7b90 100644 --- a/sky/backends/cloud_vm_ray_backend.py +++ b/sky/backends/cloud_vm_ray_backend.py @@ -50,6 +50,7 @@ from sky.skylet import job_lib from sky.skylet import log_lib from sky.usage import us...
{ "difficulty": "medium", "estimated_review_effort": 2, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-2921@9a58373
skypilot-org/skypilot
Python
2,921
[Core/AWS] Skip association with no subnetId and fix autodown for spot instances
<!-- Describe the changes in this PR --> Fixes #2920. <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the relevant ones): - [ ] Code formatting: `bash format.sh` - [ ] Any manual or new tests for this PR (pl...
2023-12-29T15:59:41Z
[AWS] KeyError 'SubnetId' when multiple Main route tables appear in a same region `sky launch --cloud aws --region us-west-2 --cpus 2 echo hi` ``` E 12-29 15:55:13 provisioner.py:88] Failed to configure the cloud for 'test-west' with the following error: E 12-29 15:55:13 provisioner.py:88] KeyError: 'SubnetId' W 12...
[ { "body": "`sky launch --cloud aws --region us-west-2 --cpus 2 echo hi`\r\n```\r\nE 12-29 15:55:13 provisioner.py:88] Failed to configure the cloud for 'test-west' with the following error:\r\nE 12-29 15:55:13 provisioner.py:88] KeyError: 'SubnetId'\r\nW 12-29 15:55:14 cloud_vm_ray_backend.py:2246] sky.exceptio...
b53976a4ab8be34bc7a398854924632e3f52b812
{ "head_commit": "9a583738e9356675f5a8922fa59b71487282ebac", "head_commit_message": "Skip association if no subnetid", "patch_to_review": "diff --git a/sky/provision/aws/config.py b/sky/provision/aws/config.py\nindex b03835068f0..b113b1a5714 100644\n--- a/sky/provision/aws/config.py\n+++ b/sky/provision/aws/confi...
[ { "diff_hunk": "@@ -218,7 +218,10 @@ def _is_subnet_public(ec2, subnet_id, vpc_id: Optional[str]) -> bool:\n rt for rt in all_route_tables\n # An RT can be associated with multiple subnets, i.e.,\n # rt['Associations'] is a list of associations.\n- if any(assoc['SubnetId'] == subn...
c69321eae394872cf2cf617b22ea7af7be45145b
diff --git a/sky/backends/cloud_vm_ray_backend.py b/sky/backends/cloud_vm_ray_backend.py index 1f0d9c23452..ac9e9e11d40 100644 --- a/sky/backends/cloud_vm_ray_backend.py +++ b/sky/backends/cloud_vm_ray_backend.py @@ -4277,21 +4277,13 @@ def set_autostop(self, idle_minutes_to_autostop: Optional[int...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-3157@4cd2ea9
skypilot-org/skypilot
Python
3,157
[Core] Optimize kubernetes cmd executions with kubernetes command runner
<!-- Describe the changes in this PR --> Fixes #3154, by refactoring the command runner and avoid using ssh in the backend for kubernetes pods. This is helpful for the future support of Slurm and other services that do not have ssh enabled. This is also a prototype to reduce the launch and exec time on kubernetes...
2024-02-14T18:55:20Z
[k8s] Slow `file_mounts` on k8s Opening a SSH connection on k8s is high latency because of two extra steps: 1. Goes through k8s API server (`kubectl port-forward`) 2. Jumps through SSH jump pod. This makes many operations slow, including file_mounts. Example, mounting three 1kb files takes 22s on EKS K8s compare...
[ { "body": "Opening a SSH connection on k8s is high latency because of two extra steps:\r\n\r\n1. Goes through k8s API server (`kubectl port-forward`)\r\n2. Jumps through SSH jump pod.\r\n\r\nThis makes many operations slow, including file_mounts. Example, mounting three 1kb files takes 22s on EKS K8s compared t...
26d902d7e47900bb6b6c897f6fda79047b35df35
{ "head_commit": "4cd2ea9eae8a82eb90120440b5789d4cb341a32c", "head_commit_message": "longer wait", "patch_to_review": "diff --git a/sky/backends/backend_utils.py b/sky/backends/backend_utils.py\nindex b1598c7c039..2016949f96a 100644\n--- a/sky/backends/backend_utils.py\n+++ b/sky/backends/backend_utils.py\n@@ -26...
[ { "diff_hunk": "@@ -158,6 +159,15 @@ def _raise_pod_scheduling_errors(namespace, new_nodes):\n raise config_lib.KubernetesError(f'{timeout_err_msg}')\n \n \n+def _raise_command_running_error(message: str, command: str, pod_name: str,\n+ rc: int, stdout: str):", "line": nu...
f5018f19301ab255ece8bebd1fb91f00f973e586
diff --git a/sky/backends/backend_utils.py b/sky/backends/backend_utils.py index 4d40806bf26..03f644930f4 100644 --- a/sky/backends/backend_utils.py +++ b/sky/backends/backend_utils.py @@ -2654,27 +2654,6 @@ def stop_handler(signum, frame): raise KeyboardInterrupt(exceptions.SIGTSTP_CODE) -def run_command_...
{ "difficulty": "high", "estimated_review_effort": 4, "problem_domain": "Performance Optimizations" }
skypilot-org__skypilot-2904@cffbabf
skypilot-org/skypilot
Python
2,904
[Core] Add option to avoid uploading credentials and custom labels for GCP
<!-- Describe the changes in this PR --> This is to avoid uploading credentials of the current cloud to be launched, as we can use the service account assigned to the instance to access the buckets and cloud APIs. Fixes #2820 To enable this feature, a user should specify the following config in the `~/.sky/con...
2023-12-27T08:24:53Z
Skypilot copies `.aws/credentials` Hi! We've started using skypilot recently. Thanks for the great tool. :) We noticed that when we have a `.aws/credentials` files locally on our machine, Skypilot seems to copy it to the remote. At the same time, the remote also receive permissions from the `skypilot-v1` role. ...
Thanks for the report @Wirg. We should definitely fix it. For your local machine, are you using any AWS profile (`AWS_PROFILE=xxx`)? What's the credential type you expect to use for local machine and subsequently any SkyPilot launched instances? Hi @concretevitamin On my local machine, I am using access token for...
[ { "body": "Hi!\r\n\r\nWe've started using skypilot recently. Thanks for the great tool. :)\r\n\r\nWe noticed that when we have a `.aws/credentials` files locally on our machine, Skypilot seems to copy it to the remote.\r\nAt the same time, the remote also receive permissions from the `skypilot-v1` role.\r\nWe e...
bd8c2eafaa1e2b387de4531bb3400896653a4035
{ "head_commit": "cffbabfc3ed68d0e7f588db080bd544e0d84292b", "head_commit_message": "address comments", "patch_to_review": "diff --git a/docs/source/reference/config.rst b/docs/source/reference/config.rst\nindex 87dafef38db..20a2228dc58 100644\n--- a/docs/source/reference/config.rst\n+++ b/docs/source/reference/c...
[ { "diff_hunk": "@@ -1509,6 +1509,17 @@ def _retry_zones(\n # does not have nodes labeled with GPU types.\n logger.info(f'{e}')\n continue\n+ except exceptions.InvalidCloudConfigs as e:\n+ # Failed due to invalid user input, e.g. invalid c...
a0f523499238c54882eaac4dbb8f7978d1c22c21
diff --git a/docs/source/reference/config.rst b/docs/source/reference/config.rst index 87dafef38db..9b27762761f 100644 --- a/docs/source/reference/config.rst +++ b/docs/source/reference/config.rst @@ -109,9 +109,43 @@ Available fields and semantics: # permission to create a security group. security_group_name...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-2841@f121382
skypilot-org/skypilot
Python
2,841
[Dependency] Remove AWS dependency by default
<!-- Describe the changes in this PR --> Fixes #2810 This PR removes AWS dependency from the default installation dependency. Reason: 1. AWS dependency is not actually lightweight - with AWS dependency: 39.5s, without AWS dependency: 31.1s (20% difference) 2. AWS depdency has a strict version requirement `co...
2023-12-05T04:50:28Z
[Core] Don't include AWS dependencies by default When SkyPilot was first released, it was a convenience to include AWS dependencies by default upon install, and according to the comment in setup.py, was a bit of a "why not" because it installed quickly. However, wow there are many backends supported, including some whi...
Just adding to this issue, as there's frequent dependency issues for numerous other libraries, I'd suggest not having `awscli` in any released package. I consider best practice to use `boto3`, and would push for refactoring any existing code depending on `awscli` to use `boto3`, as the former is built on top of the lat...
[ { "body": "When SkyPilot was first released, it was a convenience to include AWS dependencies by default upon install, and according to the comment in setup.py, was a bit of a \"why not\" because it installed quickly. However, wow there are many backends supported, including some which don't use a cloud provide...
7609e229a448a8f7d27642a8d34d33b6e0534702
{ "head_commit": "f121382f1bdfcedb9aa803a9ef4f6178761ad52c", "head_commit_message": "Fix aws check credentials", "patch_to_review": "diff --git a/docs/source/getting-started/installation.rst b/docs/source/getting-started/installation.rst\nindex 70b6f9c4ab6..464cda0a99d 100644\n--- a/docs/source/getting-started/in...
[ { "diff_hunk": "@@ -23,7 +23,7 @@ Install SkyPilot using pip:\n \n pip install skypilot-nightly\n \n- # Or, choose an extra (defaults to [aws]):\n+ # Or, choose an extra:", "line": null, "original_line": 26, "original_start_line": null, "path": "docs/source/getting-st...
eae0ee909e190d2f369b0d82217d1d89128e7ebc
diff --git a/docs/source/getting-started/installation.rst b/docs/source/getting-started/installation.rst index 70b6f9c4ab6..b94c054f907 100644 --- a/docs/source/getting-started/installation.rst +++ b/docs/source/getting-started/installation.rst @@ -16,22 +16,21 @@ Install SkyPilot using pip: .. code-block:: ...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Dependency Updates & Env Compatibility" }
skypilot-org__skypilot-2645@95f42c7
skypilot-org/skypilot
Python
2,645
Implement config validation
<!-- Describe the changes in this PR --> Closes #2474 Implements validation for the `~/.sky/config.yaml` file. If there is an error, an exception is thrown and the user is warned. To avoid circular imports, I had to refactor the code a bit, moving validation code into `common_utils` and moving the imports in ...
2023-10-03T09:23:35Z
[config] Pre-checks for `~/.sky/config.yaml` Currently, we don't have a way to pre-check the `~/.sky/config.yaml`, making it very easy to miss spell or setting wrong configs in the yaml. We can apply the json schema check to the `~/.sky/config.yaml` to make sure there is no unexpected settings. A user met the issue ...
Can I pick this up? #2600 ran into this. Let me know if you're still interested in picking this up @djokester! Happy to guide you.
[ { "body": "Currently, we don't have a way to pre-check the `~/.sky/config.yaml`, making it very easy to miss spell or setting wrong configs in the yaml. We can apply the json schema check to the `~/.sky/config.yaml` to make sure there is no unexpected settings.\r\n\r\nA user met the issue when specifying the sp...
470db0c2a85862034a24e9a1b839d9624de4bbc7
{ "head_commit": "95f42c7b09108f4c166eb935cab2fcfe289425a4", "head_commit_message": "Add test for invalid enum in config", "patch_to_review": "diff --git a/sky/authentication.py b/sky/authentication.py\nindex 442c1abb372..4ab88fe28a4 100644\n--- a/sky/authentication.py\n+++ b/sky/authentication.py\n@@ -42,6 +42,7...
[ { "diff_hunk": "@@ -160,6 +161,11 @@ def _try_load_config() -> None:\n logger.debug(f'Config loaded:\\n{pprint.pformat(_dict)}')\n except yaml.YAMLError as e:\n logger.error(f'Error in loading config file ({config_path}):', e)\n+ if _dict is not None:\n+ common_...
808bbf27d008e5143e164f15fa4cdbc769a8308b
diff --git a/docs/source/reference/config.rst b/docs/source/reference/config.rst index 43f34a48483..5c0a86ec66b 100644 --- a/docs/source/reference/config.rst +++ b/docs/source/reference/config.rst @@ -113,3 +113,33 @@ Available fields and semantics: # Only one element is allowed in this list, as GCP disallows mu...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "Code Refactoring / Architectural Improvement" }
skypilot-org__skypilot-2767@dd4f2fd
skypilot-org/skypilot
Python
2,767
SkyBenchmark: fix job_status is None for failed candidates.
<!-- Describe the changes in this PR --> Attempt to fix https://github.com/skypilot-org/skypilot/issues/2765. Repro: ``` » cat test.yaml run: | nvidia-smi resources: cloud: gcp region: us-central1 zone: us-central1-f sky bench launch test.yaml --gpus A100:8,T4 --benchmark test1 ``` Before...
2023-11-10T17:19:02Z
[Benchmark] TypeError: '<' not supported between instances of 'NoneType' and 'JobStatus' I created a benchmark that included a candidate `- {accelerators: A100:8}`. ```bash sky bench launch benchmark.yaml --benchmark hydra-3eps ``` At the time, no nodes were available in GCP that could satisfy that candidate, s...
I'm not sure if it's related to the failed candidate, but it also appears that all of the benchmark clusters were kept running throughout the entire benchmark(even though some should have definitely completed sooner). They all have a duration of exactly `2h 2m 2s` Thanks for the report @csaroff. Could you help test if ...
[ { "body": "I created a benchmark that included a candidate `- {accelerators: A100:8}`.\r\n\r\n```bash\r\nsky bench launch benchmark.yaml --benchmark hydra-3eps\r\n```\r\n\r\nAt the time, no nodes were available in GCP that could satisfy that candidate, so it wasn't provisioned.\r\n\r\nUnfortunately, when I went...
feacc9f28e2f39220bb3e79a6085924a9250426a
{ "head_commit": "dd4f2fd9be121040eb77b7c03e4916c38897a506", "head_commit_message": "Fix --gpus parsing", "patch_to_review": "diff --git a/sky/benchmark/benchmark_utils.py b/sky/benchmark/benchmark_utils.py\nindex 7db60783747..80cd3692398 100644\n--- a/sky/benchmark/benchmark_utils.py\n+++ b/sky/benchmark/benchma...
[ { "diff_hunk": "@@ -321,8 +321,10 @@ def _update_benchmark_result(benchmark_result: Dict[str, Any]) -> Optional[str]:\n stream_logs=False)[1]\n \n # Update the benchmark status.\n- if (cluster_status == status_lib.ClusterStatus.INIT or\n- job_sta...
5334347659398e89e32ac776a46482319398ccf2
diff --git a/sky/benchmark/benchmark_utils.py b/sky/benchmark/benchmark_utils.py index 7db60783747..3ba1f36a521 100644 --- a/sky/benchmark/benchmark_utils.py +++ b/sky/benchmark/benchmark_utils.py @@ -301,6 +301,11 @@ def _update_benchmark_result(benchmark_result: Dict[str, Any]) -> Optional[str]: run_end_path = o...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-2636@ca0d6ad
skypilot-org/skypilot
Python
2,636
[k8s] Fix GPU count detection from autoscaler YAML
Closes #2608. Our resource detection from pod spec puts a minimum of 1 on CPU resource so that fractional CPU sized clusters can also be launched on k8s clusters. However, this was being incorrectly applied to GPU resources too, causing all clusters to be initialized with 1 GPU resource. Tested (run the relevant...
2023-10-02T03:03:41Z
[k8s] Ray cluster is incorrectly initialized with `gpu` resource On a CPU-only kubernetes cluster, the ray cluster shows 1 GPU available, which is incorrect. Ssh into node and run `ray status`: ``` (base) sky@test-2ea4-ray-head:~$ ray status ======== Autoscaler status: 2023-09-25 23:08:16.304098 ======== Node statu...
[ { "body": "On a CPU-only kubernetes cluster, the ray cluster shows 1 GPU available, which is incorrect. Ssh into node and run `ray status`:\r\n```\r\n(base) sky@test-2ea4-ray-head:~$ ray status\r\n======== Autoscaler status: 2023-09-25 23:08:16.304098 ========\r\nNode status\r\n---------------------------------...
a33c9dc4d4b21d67928a0f00dcc9e59fafdb3fd6
{ "head_commit": "ca0d6ad606de7b54bdd929732992e047bedfbecd", "head_commit_message": "fix gpu count detection in yaml", "patch_to_review": "diff --git a/sky/skylet/providers/kubernetes/config.py b/sky/skylet/providers/kubernetes/config.py\nindex 2ab466f03ed..f4c1623f1a2 100644\n--- a/sky/skylet/providers/kubernete...
[ { "diff_hunk": "@@ -128,11 +128,16 @@ def get_resource(container_resources: Dict[str, Any],\n # float('inf') means there's no limit set\n res_count = request if limit == float('inf') else limit\n # Convert to int since Ray autoscaler expects int.\n- # Cap the minimum resource to 1 because if reso...
c10208d3513a47fbe1f591575166f0dbf780e9a5
diff --git a/sky/skylet/providers/kubernetes/config.py b/sky/skylet/providers/kubernetes/config.py index 2ab466f03ed..cfc08523713 100644 --- a/sky/skylet/providers/kubernetes/config.py +++ b/sky/skylet/providers/kubernetes/config.py @@ -128,11 +128,17 @@ def get_resource(container_resources: Dict[str, Any], # floa...
{ "difficulty": "low", "estimated_review_effort": 2, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-2491@8495134
skypilot-org/skypilot
Python
2,491
[Core] Fix status refresh for INIT cluster
<!-- Describe the changes in this PR --> Fixes #2490 <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the relevant ones): - [x] Code formatting: `bash format.sh` - [x] Any manual or new tests for this PR (pl...
2023-08-30T20:57:03Z
[Core] Assertion error when refreshing status for INIT cluster To reproduce: ``` sky launch -c existing-up-cluster # ctrl-c when it is launching sky status -r existing-up-cluster sky launch -c existing-up-cluster ``` ``` sky/backends/backend_utils.py", line 2024, in run_ray_status_to_check_ray_cluster_healthy...
[ { "body": "To reproduce:\r\n```\r\nsky launch -c existing-up-cluster\r\n# ctrl-c when it is launching\r\nsky status -r existing-up-cluster\r\nsky launch -c existing-up-cluster\r\n```\r\n\r\n```\r\nsky/backends/backend_utils.py\", line 2024, in run_ray_status_to_check_ray_cluster_healthy\r\n assert handle.hea...
eede3715ea67933bdfd2530d4fd6e37be825b871
{ "head_commit": "8495134f7e04676c0b6fb8bd19a11d265df95ed2", "head_commit_message": "changed to runtime error", "patch_to_review": "diff --git a/sky/backends/backend_utils.py b/sky/backends/backend_utils.py\nindex 64564f3eb0d..410e3df9a1c 100644\n--- a/sky/backends/backend_utils.py\n+++ b/sky/backends/backend_uti...
[ { "diff_hunk": "@@ -2014,14 +2014,22 @@ def run_ray_status_to_check_ray_cluster_healthy() -> bool:\n # triggered.\n if external_ips is None or len(external_ips) == 0:\n logger.debug(f'Refreshing status ({cluster_name!r}): No cached '\n- f'IPs f...
6f75c0671325168990db5795764a863a7316d7cd
diff --git a/sky/backends/backend_utils.py b/sky/backends/backend_utils.py index 16bd01e39d9..33d7ab26daa 100644 --- a/sky/backends/backend_utils.py +++ b/sky/backends/backend_utils.py @@ -2012,17 +2012,23 @@ def run_ray_status_to_check_ray_cluster_healthy() -> bool: # triggered. if external_i...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-2403@c84c1ff
skypilot-org/skypilot
Python
2,403
[Core] Auto mapping for cluster name
<!-- Describe the changes in this PR --> Closes #1853 and Closes #963 The backward compatibility is guaranteed by the restore of ray yaml in the `backend_utils.write_cluster_config` <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the c...
2023-08-14T22:06:55Z
Gracefully handle long cluster names & for spot, show FAILED_CONTROLLER instead Been failing test_inline_spot_env smoke test: ``` » sky spot status | grep test-inline-spot-env-zongheng-4edc-62 428 test-inline-spot-env-zongheng-4edc-62 1x [CPU:0.5] 4 mins ago 2m 27s - 0 FAILED_NO_...
`test_id` was added to avoid the second smoke test reusing the cluster launched in the first smoke test and forgot to be terminated. Also, for the `test_spot_recovery`, we have to have that, to make sure the manual termination with `aws ec2` does not accidentally terminate other spot clusters from the different `spot l...
[ { "body": "Been failing test_inline_spot_env smoke test:\r\n```\r\n» sky spot status | grep test-inline-spot-env-zongheng-4edc-62\r\n428 test-inline-spot-env-zongheng-4edc-62 1x [CPU:0.5] 4 mins ago 2m 27s - 0 FAILED_NO_RESOURCE\r\n\r\n» sky logs sky-spot-controller 428\r\nI...
469a62d9ac20043f6d267edcf04f5928dae65034
{ "head_commit": "c84c1ff3d0a23f88e37f64a0f4c754dbb0da6580", "head_commit_message": "Add a readme", "patch_to_review": "diff --git a/sky/backends/backend_utils.py b/sky/backends/backend_utils.py\nindex 2ccc9235198..1c45491c8dc 100644\n--- a/sky/backends/backend_utils.py\n+++ b/sky/backends/backend_utils.py\n@@ -8...
[ { "diff_hunk": "", "line": null, "original_line": null, "original_start_line": null, "path": "sky/templates/aws-ray.yml.j2", "start_line": null, "text": "@user1:\ntag head/workers?\n\n@author:\nRemoved." }, { "diff_hunk": "@@ -63,6 +63,7 @@ available_node_types:\n node_config...
2c7f159a34900d8b874151f5a9dee304db3019b6
diff --git a/sky/backends/backend_utils.py b/sky/backends/backend_utils.py index 9c39883c59f..1d793496cc4 100644 --- a/sky/backends/backend_utils.py +++ b/sky/backends/backend_utils.py @@ -892,6 +892,7 @@ def write_cluster_config( # task.best_resources may not be equal to to_provision if the user # is running...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-2729@efdd07d
skypilot-org/skypilot
Python
2,729
[K8s] Custom Image Support for Kubernetes Instances
<!-- Describe the changes in this PR --> This PR resolves #2599 When users opt to provide their custom images for setting up the k8s instance, instead of relying on Skypilot's default image built with `Dockerfile_k8s`, certain assumptions and adjustments are made to ensure smooth operation: **Base Distribution:...
2023-10-23T01:29:54Z
[k8s] Custom image support for Kubernetes We need to support custom images for Kubernetes clusters, possibly set via `image_id: docker:<image_id>`. We'd need to take care of a few things/refactor our k8s provisioning: 1. Install and configure openssh as a part of setup to ensure ssh access 2. Make sure basic de...
This issue is stale because it has been open 120 days with no activity. Remove stale label or comment or this will be closed in 10 days.
[ { "body": "We need to support custom images for Kubernetes clusters, possibly set via `image_id: docker:<image_id>`. \r\n\r\nWe'd need to take care of a few things/refactor our k8s provisioning:\r\n\r\n1. Install and configure openssh as a part of setup to ensure ssh access\r\n2. Make sure basic dependencies ar...
6fb19d359cf9b54f362adea0c1a8ce5dbfe9c5c6
{ "head_commit": "efdd07deb7943666d67f000f6e09701e5161ecae", "head_commit_message": "multinode fixes", "patch_to_review": "diff --git a/Dockerfile_k8s b/Dockerfile_k8s\nindex cde67cac210..6cbc167265e 100644\n--- a/Dockerfile_k8s\n+++ b/Dockerfile_k8s\n@@ -25,26 +25,21 @@ RUN useradd -m -s /bin/bash sky && \\\n # ...
[ { "diff_hunk": "@@ -288,17 +280,19 @@ setup_commands:\n # Line 'sudo grep ..': set the number of threads per process to unlimited to avoid ray job submit stucking issue when the number of running ray jobs increase.\n # Line 'mkdir -p ..': disable host key check\n # Line 'python3 -c ..': patch the buggy ra...
2b842b98a6578ecc310caf081e5d2fe79193f1f0
diff --git a/sky/skylet/providers/kubernetes/node_provider.py b/sky/skylet/providers/kubernetes/node_provider.py index 2e58262269a..b08e6df61d1 100644 --- a/sky/skylet/providers/kubernetes/node_provider.py +++ b/sky/skylet/providers/kubernetes/node_provider.py @@ -1,5 +1,7 @@ import copy import logging +import os +im...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "New Feature Additions" }
skypilot-org__skypilot-2389@b12cbb6
skypilot-org/skypilot
Python
2,389
[UX] Better logging message for operators on the clusters terminated manually in the cloud
<!-- Describe the changes in this PR --> Fixes #2298. Note this only works for the spot clusters or the clusters with autostop/down setup, as the status refresh will not be triggered for the operators on the normal clusters. Previously: ``` sky logs test-warning-spot ValueError: Cluster 'test-warning-spot' ...
2023-08-11T18:45:42Z
Ops with auto-refresh should warn about clusters terminated on the cloud Currently, clusters terminated on the cloud (either manually, or preempted, or auto-downed) are only warned if they are caught in `status -r`, where we print "Cluster terminated on the cloud: xxx". However, for the preemption case, user has run...
[ { "body": "Currently, clusters terminated on the cloud (either manually, or preempted, or auto-downed) are only warned if they are caught in `status -r`, where we print \"Cluster terminated on the cloud: xxx\".\r\n\r\nHowever, for the preemption case, user has run into the following:\r\n- a preemption happened\...
63a99eb927fdd0c0b0a784c36cf3e5903427e7b8
{ "head_commit": "b12cbb661ed45d9f3579acec6ec3d7837598ce07", "head_commit_message": "Fix message", "patch_to_review": "diff --git a/sky/backends/backend_utils.py b/sky/backends/backend_utils.py\nindex 4e734f7e0fb..c711aec3374 100644\n--- a/sky/backends/backend_utils.py\n+++ b/sky/backends/backend_utils.py\n@@ -23...
[ { "diff_hunk": "@@ -2390,10 +2394,15 @@ def check_cluster_available(\n bright = colorama.Style.BRIGHT\n reset = colorama.Style.RESET_ALL\n if handle is None:\n+ if previous_cluster_status is None:\n+ error_msg = f'Cluster {cluster_name!r} does not exist.'\n+ else:\n+ ...
c7ee0c8acea2e404eba36448dddae6564eddaf4c
diff --git a/sky/backends/backend_utils.py b/sky/backends/backend_utils.py index 081cf3df1bf..0ec5d0ecb07 100644 --- a/sky/backends/backend_utils.py +++ b/sky/backends/backend_utils.py @@ -2338,10 +2338,15 @@ def check_cluster_available( exceptions.CloudUserIdentityError: if we fail to get the current user ...
{ "difficulty": "low", "estimated_review_effort": 2, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-2804@35b7257
skypilot-org/skypilot
Python
2,804
[Storage] Error Handle Attempts to Mount Non-sky Managed Bucket without Source
<!-- Describe the changes in this PR --> This closes #2779, #2950 `Storage` is designed in a way to not allow externally created cloud storages to be mounted without specifying the `source` field. The only way to mount non-sky managed bucket is to specify the bucket's URI as the `source` field. This fix raises `Sto...
2023-11-19T09:53:42Z
[Storage] sky start failing to re-mount on externally created Storage User reported to @Michaelvll : > Got this error while restarting an instance: > `sky.exceptions.StorageExternalDeletionError: The bucket, 'bucket-name', could not be mounted on cluster 'cluster-name'. Please verify that the bucket exists. The clu...
[ { "body": "User reported to @Michaelvll :\r\n\r\n> Got this error while restarting an instance:\r\n> `sky.exceptions.StorageExternalDeletionError: The bucket, 'bucket-name', could not be mounted on cluster 'cluster-name'. Please verify that the bucket exists. The cluster started successfully without mounting th...
5cc68d2c9be0870d23b5e73ded1484b2e851268e
{ "head_commit": "35b7257aedae210711027528cd0d2d00a422be43", "head_commit_message": "add new smoke test", "patch_to_review": "diff --git a/sky/data/storage.py b/sky/data/storage.py\nindex d9f8156fa89..89dc67ade3f 100644\n--- a/sky/data/storage.py\n+++ b/sky/data/storage.py\n@@ -819,6 +819,10 @@ def add_store(self...
[ { "diff_hunk": "@@ -1248,6 +1254,26 @@ def _get_bucket(self) -> Tuple[StorageHandle, bool]:\n # bucket or if it is a user's bucket that is publicly\n # accessible.\n self.client.head_bucket(Bucket=self.name)\n+ # Check if 'source' is None, indicating Storage is in ...
dd1f717c344df0229131ae55c3e488c1b63e04b8
diff --git a/sky/data/storage.py b/sky/data/storage.py index 065f5aeb1ff..4ca8441be3a 100644 --- a/sky/data/storage.py +++ b/sky/data/storage.py @@ -317,6 +317,31 @@ def __deepcopy__(self, memo): # original Store object is returned return self + def _validate_existing_bucket(self): + """Va...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-2296@684c794
skypilot-org/skypilot
Python
2,296
Add support for env files.
<!-- Describe the changes in this PR --> Fixes #2295 A new argument is added `--env-file`. This reads from a dotenv file to read multiple environment variables. These are merged with any values passed via `--env` arguments with `--env` taking precedence in any conflicts. <!-- Describe the tests ran --> No...
2023-07-24T06:48:15Z
[Feature Request] Support passing env variables via a .env file When launching or starting a new job, env variables currently need to be embedded in the config file or passed in through multiple `--env` arguments. In some setups, its more convenient to have multiple env variables in a `.env` file. Much like Docker's ...
Thanks for the great request @fozziethebeat! We got a similar request in #2151 as well. We will increase the priority for this. : ) Ah yeah that's fairly similar. When I use workdir, copying over the env file works well. But sometimes I want to skip workdir since my workdir can be a mess. Good point! We are happy to ...
[ { "body": "When launching or starting a new job, env variables currently need to be embedded in the config file or passed in through multiple `--env` arguments. In some setups, its more convenient to have multiple env variables in a `.env` file. Much like Docker's support, it'd be nice to be able to do \r\n\r...
f5526d43b49f5e8bf184f73c58193d9d203cdc1e
{ "head_commit": "684c794fd915904a89bbe4dc264feee2c5a02c90", "head_commit_message": "Add support for env files. Env args will override any env file values and be passed on to override any yaml values", "patch_to_review": "diff --git a/sky/cli.py b/sky/cli.py\nindex 3136a16678a..5dd15e24c8d 100644\n--- a/sky/cli....
[ { "diff_hunk": "@@ -86,6 +86,7 @@ def parse_readme(readme: str) -> str:\n # PrettyTable with version >=2.0.0 is required for the support of\n # `add_rows` method.\n 'PrettyTable>=2.0.0',\n+ 'python-dotenv==1.0.0',", "line": null, "original_line": 89, "original_start_line": null, "...
5ed5a2da9a4e04f0036961309028eb957f702a9c
diff --git a/examples/sample_dotenv b/examples/sample_dotenv new file mode 100644 index 00000000000..ac56cc1c950 --- /dev/null +++ b/examples/sample_dotenv @@ -0,0 +1 @@ +TEST_ENV2="success" diff --git a/sky/cli.py b/sky/cli.py index 3136a16678a..1e27275d4f9 100644 --- a/sky/cli.py +++ b/sky/cli.py @@ -40,6 +40,7 @@ ...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "New Feature Additions" }
skypilot-org__skypilot-2322@d68c0fa
skypilot-org/skypilot
Python
2,322
[Storage] Support for storage info in cluster table's metadata
<!-- Describe the changes in this PR --> This PR resolves #1203. Currently, when we run `sky stop <cluster-name>` on a cluster that has cloud storage mounted, the mounting will be lost when restarted with `sky start <cluster-name>` as the mounting command does not get ran. In this PR, I save the `storage_mounts...
2023-07-28T23:40:24Z
[Storage] VM loses S3 mount after stop/start To reproduce: make `cluster.yml`: ``` resources: cloud: aws instance_type: g4dn.xlarge disk_size: 500 workdir: . num_nodes: 4 file_mounts: /dataset: source: s3://raysort-kl store: s3 setup: | echo "Hello, SkyPilot!" run: | echo ...
Yes... current `start` will not mount the s3 bucket again. Please try to run ~~`sky launch --no-setup -y`~~ `sky launch cluster.yml -c g4 --no-setup`. This problem was recently brought up by @lhqing as well, in #1157. TODO: Add the `sky launch --no-setup` in the doc for the storage mount. Thanks! Just to make sure,...
[ { "body": "To reproduce: make `cluster.yml`:\r\n\r\n```\r\nresources:\r\n cloud: aws\r\n instance_type: g4dn.xlarge\r\n disk_size: 500\r\n\r\nworkdir: .\r\n\r\nnum_nodes: 4\r\n\r\nfile_mounts:\r\n /dataset:\r\n source: s3://raysort-kl\r\n store: s3\r\n\r\nsetup: |\r\n echo \"Hello, SkyPilot!\"\r\n\r\...
9ff192705679eda2447a2fa1642c0392420942b8
{ "head_commit": "d68c0fab6ad56cbb52f135f0f3ee00c7ed4d487c", "head_commit_message": "format", "patch_to_review": "diff --git a/sky/backends/cloud_vm_ray_backend.py b/sky/backends/cloud_vm_ray_backend.py\nindex 941a60ec299..f1e57f52198 100644\n--- a/sky/backends/cloud_vm_ray_backend.py\n+++ b/sky/backends/cloud_vm...
[ { "diff_hunk": "@@ -4240,6 +4241,45 @@ def _execute_storage_mounts(self, handle: CloudVmRayResourceHandle,\n end = time.time()\n logger.debug(f'Storage mount sync took {end - start} seconds.')\n \n+ def _set_cluster_metadata_storage_mounts(\n+ self, handle: CloudVmRayResourceHandle...
16efcd74ca7793ec7d0330a44abb93f04c19c34b
diff --git a/sky/backends/backend.py b/sky/backends/backend.py index 28aa981b078..c51ecd41d92 100644 --- a/sky/backends/backend.py +++ b/sky/backends/backend.py @@ -66,8 +66,8 @@ def sync_workdir(self, handle: _ResourceHandleType, workdir: Path) -> None: def sync_file_mounts( self, handle: _Resou...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-2279@0e957e4
skypilot-org/skypilot
Python
2,279
[GCP] Use the latest debian image to align python 3.10
<!-- Describe the changes in this PR --> Fixes #1604 <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the relevant ones): - [x] Code formatting: `bash format.sh` - [x] Any manual or new tests for this PR (pl...
2023-07-20T04:37:16Z
Python version mismatch when migrating from AWS to GCP # Description of Problem I recently migrated my workload from AWS to GCP and ran into issues with python versions. AWS's deep learning AMI is based on LTS versions of Ubuntu with Python 3.8+, whereas the default SkyPilot GCP image uses Debian 10 with Python 3.7. T...
I have similar problems with the python versions, so I am following up here. Several tools I use have bumped their minimum python version to py3.8, while the GCP sky-VM default python is still py3.7. I can do a similar workaround like @pschafhalter wrote above. I wonder the SkyPilot has some built-in solutions for ...
[ { "body": "# Description of Problem\r\nI recently migrated my workload from AWS to GCP and ran into issues with python versions. AWS's deep learning AMI is based on LTS versions of Ubuntu with Python 3.8+, whereas the default SkyPilot GCP image uses Debian 10 with Python 3.7. This resulted in an error as some l...
ca6d1e841875df290667cee02082ae7e2149126a
{ "head_commit": "0e957e4bc40dcecea76bcbb340ade6e35765ba4b", "head_commit_message": "Merge branch 'master' of github.com:skypilot-org/skypilot into align-py310-gcp", "patch_to_review": "diff --git a/sky/clouds/gcp.py b/sky/clouds/gcp.py\nindex a6d32751928..1b2d395ee83 100644\n--- a/sky/clouds/gcp.py\n+++ b/sky/cl...
[ { "diff_hunk": "@@ -27,6 +27,11 @@\n _image_df = common.read_catalog('gcp/images.csv',\n pull_frequency_hours=_PULL_FREQUENCY_HOURS)\n \n+if _image_df[_image_df['Tag'] == 'skypilot:gpu-debian-11'].empty:", "line": null, "original_line": 30, "original_start_line": null...
880d967860a6d757c3e8a6949b154debaee6b782
diff --git a/sky/clouds/gcp.py b/sky/clouds/gcp.py index a6d32751928..1b2d395ee83 100644 --- a/sky/clouds/gcp.py +++ b/sky/clouds/gcp.py @@ -344,7 +344,7 @@ def make_deploy_resources_variables( # --no-standard-images # We use the debian image, as the ubuntu image has some connectivity # issue...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Dependency Updates & Env Compatibility" }
skypilot-org__skypilot-2265@99c0906
skypilot-org/skypilot
Python
2,265
Fix Lambda rate limit by backoff and retry when exceeding
<!-- Describe the changes in this PR --> This PR fixes #2261 by: * Add retry with backoff to all Lambda API requests when the rate limit exceeds; * Retry launching when exceeding the rate limit. TODO: debug and verify it works properly <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are pa...
2023-07-18T20:25:45Z
[Lambda] Handle rate limit error crashing `sky launch` User reported: > For lambda, I'm getting api rate limit errors. > However, the problem is that I'm guessing there are some secondary api calls that causes sky launch to fail because of the rate limi, so the job never gets sent, but the node itself gets provisio...
Can reproduce this with: ``` for i in $(seq 5); do sky launch --cpus 1+ --cloud lambda --down -y -c lambda-$i &; done ```
[ { "body": "User reported:\r\n> For lambda, I'm getting api rate limit errors. \r\n> However, the problem is that I'm guessing there are some secondary api calls that causes sky launch to fail because of the rate limi, so the job never gets sent, but the node itself gets provisioned. \r\n> And because sky launch...
f5526d43b49f5e8bf184f73c58193d9d203cdc1e
{ "head_commit": "99c090671ec348489ea46396fe83630f6ef852a1", "head_commit_message": "merge to master", "patch_to_review": "diff --git a/sky/backends/cloud_vm_ray_backend.py b/sky/backends/cloud_vm_ray_backend.py\nindex d3dd34db2b4..5b28cd43660 100644\n--- a/sky/backends/cloud_vm_ray_backend.py\n+++ b/sky/backends...
[ { "diff_hunk": "@@ -2,10 +2,17 @@\n import json\n import os\n import requests\n+import time\n from typing import Any, Dict, List, Optional, Tuple\n \n+from sky.utils import common_utils\n+\n CREDENTIALS_PATH = '~/.lambda_cloud/lambda_keys'\n API_ENDPOINT = 'https://cloud.lambdalabs.com/api/v1'\n+# TODO(tian): D...
911bd5bbef75abca0c2fe3da5566a5e3dbb45d97
diff --git a/sky/backends/cloud_vm_ray_backend.py b/sky/backends/cloud_vm_ray_backend.py index d3dd34db2b4..5b28cd43660 100644 --- a/sky/backends/cloud_vm_ray_backend.py +++ b/sky/backends/cloud_vm_ray_backend.py @@ -1787,6 +1787,12 @@ def need_ray_up( 'error.') return True...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-2262@edc2ffa
skypilot-org/skypilot
Python
2,262
Fix gcp t4 attach instance
<!-- Describe the changes in this PR --> Fixes #2038. <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the relevant ones): - [x] Code formatting: `bash format.sh` - [x] Any manual or new tests for this PR (pl...
2023-07-18T06:32:25Z
GCP logic paired `T4:4` with a wrong instance type ``` » sky launch --gpus T4:4 --cloud gcp I 06-05 11:00:37 optimizer.py:636] == Optimizer == I 06-05 11:00:37 optimizer.py:648] Target: minimizing cost I 06-05 11:00:37 optimizer.py:659] Estimated cost: $4.4 / hour I 06-05 11:00:37 optimizer.py:659] ...
[ { "body": "```\r\n» sky launch --gpus T4:4 --cloud gcp \r\nI 06-05 11:00:37 optimizer.py:636] == Optimizer ==\r\nI 06-05 11:00:37 optimizer.py:648] Target: minimizing cost\r\nI 06-05 11:00:37 optimizer.py:659] Estimated cost: $4.4 / hour\r\nI 06-05 11:00:37 optimizer.py:659]\r\nI 06-05 11:00:37 opt...
a184687490b6c596dc7b76a0ba9fc4728012f58c
{ "head_commit": "edc2ffaaeea4fef69cd5cc07e1052537519cd18d", "head_commit_message": "add comment", "patch_to_review": "diff --git a/sky/clouds/service_catalog/gcp_catalog.py b/sky/clouds/service_catalog/gcp_catalog.py\nindex e3f32a8ceef..5d5f3586787 100644\n--- a/sky/clouds/service_catalog/gcp_catalog.py\n+++ b/s...
[ { "diff_hunk": "@@ -249,14 +253,17 @@ def get_instance_type_for_accelerator(\n if acc_name not in _NUM_ACC_TO_NUM_CPU:\n acc_name = 'DEFAULT'\n \n- assert _DEFAULT_HOST_VM_FAMILY == 'n1'\n-\n+ default_host_cpus = _NUM_ACC_TO_NUM_CPU[acc_name].get(acc_count, None)\n if cpus is None and memo...
76aebaaf6143ba32ee1f49e3bb523755015e2dc2
diff --git a/sky/clouds/service_catalog/gcp_catalog.py b/sky/clouds/service_catalog/gcp_catalog.py index ddf2aeebb19..bc2520af032 100644 --- a/sky/clouds/service_catalog/gcp_catalog.py +++ b/sky/clouds/service_catalog/gcp_catalog.py @@ -48,12 +48,15 @@ # Memory: 1 GiB RAM per 1 vCPU; 'n2-highcpu', ] +# n2 is...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-2165@282c49d
skypilot-org/skypilot
Python
2,165
[Storage] gsutil rsync failing to sync source with dangling symlink
This PR closes #2147, which is initially raised by #2127 `aws s3 sync` is a command ran to sync Sky Storages set with `s3`/`r2` storages. When the command sees a dangling symlink, it simply ignores and continue to sync other contents. However, `gsutil rsync`, which is used to set `gcs`, does not support this and fai...
2023-07-01T19:48:45Z
[Storage] GCS syncing fails when the src directory contains courrupted symlink User reported that gcs sync errors out when the src directory contains dangling symlink. The command we run to sync the contents from local directory to cloud storage for s3 is `aws s3 sync` and `gsutil rsync -r` for gcs. When the local d...
[ { "body": "User reported that gcs sync errors out when the src directory contains dangling symlink.\r\n\r\nThe command we run to sync the contents from local directory to cloud storage for s3 is `aws s3 sync` and `gsutil rsync -r` for gcs. When the local directory being synced contains a dangling symlink(the fi...
5402440df09a946a9b06ce1adba6f2899339b0eb
{ "head_commit": "282c49d855246cb514f60212c6c7b4a237c55c65", "head_commit_message": "update_gsutil_rsync_cmd", "patch_to_review": "diff --git a/sky/data/storage.py b/sky/data/storage.py\nindex bb2c9ab2ebb..179b907f257 100644\n--- a/sky/data/storage.py\n+++ b/sky/data/storage.py\n@@ -1507,13 +1507,13 @@ def batch_...
[ { "diff_hunk": "@@ -1507,13 +1507,13 @@ def batch_gsutil_rsync(self,\n \n def get_file_sync_command(base_dir_path, file_names):\n sync_format = '|'.join(file_names)\n- sync_command = (f'gsutil -m rsync -x \\'^(?!{sync_format}$).*\\' '\n+ sync_command = (f'gsutil -m rsyn...
122dae870d36d6ea0d11d10bfa401679ab3d8044
diff --git a/sky/cloud_stores.py b/sky/cloud_stores.py index 89f8e02cb5b..33aab44501d 100644 --- a/sky/cloud_stores.py +++ b/sky/cloud_stores.py @@ -133,7 +133,7 @@ def is_directory(self, url: str) -> bool: def make_sync_dir_command(self, source: str, destination: str) -> str: """Downloads a directory usi...
{ "difficulty": "medium", "estimated_review_effort": 2, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-2196@6af8481
skypilot-org/skypilot
Python
2,196
[UX] Make sky check more fine-grained for GCP
<!-- Describe the changes in this PR --> Fixes #2184 This PR makes the `sky check` provides information based on the status of the user's current setup. Missing dependency: ``` $ sky check Checking credentials to enable clouds for SkyPilot. AWS: enabled Azure: enabled GCP: disa...
2023-07-07T21:30:35Z
[UX] `sky check` should include fine grained error messages From a user: > .. when I ran `sky check` it said gcp was disabled--even though I have `gcloud` installed locally and connected to Google Cloud. I ran a few of the commands, but when I got to `gcloud init` I didn't want to wipe my current settings. > > A...
This is excellent feedback! Important for self-setup.
[ { "body": "From a user:\r\n\r\n> .. when I ran `sky check` it said gcp was disabled--even though I have `gcloud` installed locally and connected to Google Cloud. I ran a few of the commands, but when I got to `gcloud init` I didn't want to wipe my current settings.\r\n> \r\n> After looking at the docs, what it...
75ccbf4aa732b22705601b9f76266085df8eb14e
{ "head_commit": "6af8481a73b9ab45b75aeab2e33d559b1994c66f", "head_commit_message": "Make sky check more fine-grained for GCP", "patch_to_review": "diff --git a/sky/clouds/gcp.py b/sky/clouds/gcp.py\nindex c9d44f54cf8..82422a437a0 100644\n--- a/sky/clouds/gcp.py\n+++ b/sky/clouds/gcp.py\n@@ -114,6 +114,33 @@ clas...
[ { "diff_hunk": "@@ -512,37 +550,36 @@ def check_credentials(cls) -> Tuple[bool, Optional[str]]:\n ]:\n if not os.path.isfile(os.path.expanduser(file)):\n raise FileNotFoundError(file)\n+ except FileNotFoundError as e:\n+ return False, (\n+ ...
c76c24ed6948b1f7d706e736b58ac48247f2b1db
diff --git a/sky/clouds/gcp.py b/sky/clouds/gcp.py index c9d44f54cf8..5297ca22ff5 100644 --- a/sky/clouds/gcp.py +++ b/sky/clouds/gcp.py @@ -114,6 +114,33 @@ class GCP(clouds.Cloud): # lower limit. _MAX_CLUSTER_NAME_LEN_LIMIT = 35 + _INDENT_PREFIX = ' ' + _DEPENDENCY_HINT = ( + 'GCP tools ar...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "New Feature Additions" }
skypilot-org__skypilot-2206@c8f47b1
skypilot-org/skypilot
Python
2,206
[UX] Fix unexpected output for ctrl-c during `sky launch`
<!-- Describe the changes in this PR --> Fixes #2205. The `os.killpg` might be applied to a subprocess that was attached under the root user, due the parent process being killed first. This will cause the `PermissionError: [Errno 1] Operation not permitted` @concretevitamin Could you verify if this solve the p...
2023-07-10T05:30:29Z
Weird error messages after `ctrl-c` out a running job's log ``` sky launch --cpus 1+ --use-spot --down 'for i in $(seq 10000); do echo $i; sleep 1; done' -y ``` After the job starts running, press Ctrl-C. This will then show ``` ... <status table> KeyboardInterrupt: 130 During handling of the above exception,...
[ { "body": "```\r\nsky launch --cpus 1+ --use-spot --down 'for i in $(seq 10000); do echo $i; sleep 1; done' -y\r\n```\r\nAfter the job starts running, press Ctrl-C. This will then show\r\n```\r\n...\r\n<status table>\r\nKeyboardInterrupt: 130\r\n\r\nDuring handling of the above exception, another exception occu...
1489c43f394eb6fe1a485fa918ae53afbbe15c95
{ "head_commit": "c8f47b1dbc414267e95c07602afff59cd878378e", "head_commit_message": "[UX] Fix unexpected output for ctrl-c", "patch_to_review": "diff --git a/sky/skylet/log_lib.py b/sky/skylet/log_lib.py\nindex 5debce93264..41ca95697ff 100644\n--- a/sky/skylet/log_lib.py\n+++ b/sky/skylet/log_lib.py\n@@ -262,7 +2...
[ { "diff_hunk": "@@ -262,7 +262,12 @@ def run_with_log(\n # Send SIGINT to the process directly, otherwise, the underlying\n # process will only be killed after the python program exits,\n # causing the stream handling stuck at `readline`.\n- os.killpg(proc.pid, sig...
5fe3b79cfb59b97d8778ef6a06566de5dddac787
diff --git a/sky/skylet/log_lib.py b/sky/skylet/log_lib.py index 5debce93264..8321f21d755 100644 --- a/sky/skylet/log_lib.py +++ b/sky/skylet/log_lib.py @@ -6,7 +6,6 @@ import io import multiprocessing.pool import os -import signal import subprocess import sys import time @@ -20,6 +19,7 @@ from sky.skylet import...
{ "difficulty": "medium", "estimated_review_effort": 2, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-2168@dbe44d4
skypilot-org/skypilot
Python
2,168
[UX] Fail early from provisioning when rsync is uninstalled
<!-- Describe the changes in this PR --> This PR closes #2113 When `rsync` is not installed in the user's machine, `ray_up()` is reattempted for `_MAX_RAY_UP_RETRY` number of times without failing. As this is time consuming, this PR fails early in the provisioning step when `rsync` is not installed and gently hint...
2023-07-02T03:02:40Z
[UX] Fail early when rsync is not installed locally We should fail early and show hints about installing the rsync when it is not installed on user's laptop. Currently, when user try `sky launch`, it will just keep retrying, and the error message is hidden in the `provision.log` file.
Had the same issue while testing with `sky launch` in a newly setup container environment where `rsync` is not installed.
[ { "body": "We should fail early and show hints about installing the rsync when it is not installed on user's laptop. Currently, when user try `sky launch`, it will just keep retrying, and the error message is hidden in the `provision.log` file.", "number": 2113, "title": "[UX] Fail early when rsync is n...
5402440df09a946a9b06ce1adba6f2899339b0eb
{ "head_commit": "dbe44d484f6245f94ab0a4eb7affb256a2423681", "head_commit_message": "fail early for rsync uninstalled", "patch_to_review": "diff --git a/sky/backends/cloud_vm_ray_backend.py b/sky/backends/cloud_vm_ray_backend.py\nindex 96cbb5bff19..1f3f024692a 100644\n--- a/sky/backends/cloud_vm_ray_backend.py\n+...
[ { "diff_hunk": "@@ -1336,6 +1336,22 @@ def _retry_zones(\n prev_cluster_status: Optional[status_lib.ClusterStatus],\n ):\n \"\"\"The provision retry loop.\"\"\"\n+ # When rsync is not installed in the user's machine, Ray will\n+ # silently retry to up the node for _MAX_RAY_UP_R...
393ade6251504ea4788f17d2614f0db24904d20e
diff --git a/sky/backends/backend_utils.py b/sky/backends/backend_utils.py index ecdfb227729..18fb17032f0 100644 --- a/sky/backends/backend_utils.py +++ b/sky/backends/backend_utils.py @@ -2580,3 +2580,24 @@ def run_command_and_handle_ssh_failure(runner: command_runner.SSHCommandRunner, ...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-2146@db85fe6
skypilot-org/skypilot
Python
2,146
API/YAML: Support using env vars to define file_mounts.
<!-- Describe the changes in this PR --> ``` # Use env vars in the "name" (bucket name) and "source" (local dir/file to # upload) fields. # # Both syntaxes work: ${MY_BUCKET} and $MY_BUCKET. file_mounts: /mydir: name: ${MY_BUCKET} # Name of the bucket. store: gcs mode: MOUNT /another-dir...
2023-06-29T03:27:08Z
Use environment variable to define file mount Hi and thanks for this great project ! I am in a situation where I would like to mount some directory in a bucket which changes based on the value of an environment variable. I am unsure how to do this with SkyPilot as defining a file mount would typically require that I...
Thanks for filing this report @QLutz and welcome to SkyPilot! One solution to this could be to use `envsubst` to create a temporary yaml file that you can launch. Would that work? ``` export URL_IN_MY_ENVVAR=s3://mybucket/ envsubst < template.yaml > run.yaml sky launch run.yaml ``` Alternatively, you ca...
[ { "body": "Hi and thanks for this great project !\r\n\r\nI am in a situation where I would like to mount some directory in a bucket which changes based on the value of an environment variable. I am unsure how to do this with SkyPilot as defining a file mount would typically require that I specify the correspond...
596c1415b5039adec042594f45b342374e5e6a00
{ "head_commit": "db85fe6cd8f9fbd878265c5a730ba609f8655329", "head_commit_message": "Fixes", "patch_to_review": "diff --git a/sky/cli.py b/sky/cli.py\nindex 1fcb9786f17..eeef88243f9 100644\n--- a/sky/cli.py\n+++ b/sky/cli.py\n@@ -948,7 +948,11 @@ def _check_yaml(entrypoint: str) -> Tuple[bool, Optional[Dict[str, ...
[ { "diff_hunk": "@@ -68,6 +68,45 @@ def _is_valid_env_var(name: str) -> bool:\n return bool(re.fullmatch(_VALID_ENV_VAR_REGEX, name))\n \n \n+def _fill_in_env_vars_in_storage_config(\n+ storage_config: Dict[str, Any], task_envs: Dict[str,\n+ str])...
460b07ba85e128efe2540fc1cdde7af308521b11
diff --git a/examples/using_file_mounts_with_env_vars.yaml b/examples/using_file_mounts_with_env_vars.yaml new file mode 100644 index 00000000000..74881a2ff88 --- /dev/null +++ b/examples/using_file_mounts_with_env_vars.yaml @@ -0,0 +1,52 @@ +# Example showcasing how env vars can be used in the file_mounts section. + +...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "New Feature Additions" }
skypilot-org__skypilot-2434@e35bbd0
skypilot-org/skypilot
Python
2,434
[Provisioner] Introducing `best` disk tier
<!-- Describe the changes in this PR --> This PR introduces `best` disk tier. It will automatically select the best disk tier enabled, and compatible with clouds that do not support disk tier (e.g. Lambda); in that case, it just ignores this argument and launches normally. This is helpful when you want to use the...
2023-08-20T04:38:11Z
[Optimizer] Treat `disk_tier` as a soft-constraint? [For discussion] If my job has a `disk_tier` specified, then certain clouds get excluded from the candidate list (and failover). E.g., ``` # YAML resources: accelerators: A100:1 disk_size: 1000 disk_tier: high # sky launch task.yaml # Lambda gpu_1x_a...
[ { "body": "[For discussion] If my job has a `disk_tier` specified, then certain clouds get excluded from the candidate list (and failover). E.g.,\r\n\r\n```\r\n# YAML\r\nresources:\r\n accelerators: A100:1\r\n disk_size: 1000\r\n disk_tier: high\r\n\r\n# sky launch task.yaml\r\n# Lambda gpu_1x_a100_sxm4 is n...
3765f032824bcc3ee7f27d37edbbeded377e460d
{ "head_commit": "e35bbd0001a3fa4977d97b038fa8cd6d85883f30", "head_commit_message": "add test_jobs tests", "patch_to_review": "diff --git a/docs/source/reference/yaml-spec.rst b/docs/source/reference/yaml-spec.rst\nindex c8dc4b1ed4f..76b6b6b923b 100644\n--- a/docs/source/reference/yaml-spec.rst\n+++ b/docs/source...
[ { "diff_hunk": "@@ -953,6 +939,18 @@ def less_demanding_than(\n if not self_ports <= other_ports:\n return False\n \n+ if self.disk_tier is not None:", "line": null, "original_line": 942, "original_start_line": null, "path": "sky/resources.py", "sta...
ec195cc2d192d50d6cb376698d0e55fd19f0c978
diff --git a/docs/source/reference/yaml-spec.rst b/docs/source/reference/yaml-spec.rst index e82442cf2c0..ab3067e2b21 100644 --- a/docs/source/reference/yaml-spec.rst +++ b/docs/source/reference/yaml-spec.rst @@ -99,8 +99,9 @@ Available fields: disk_size: 256 # Disk tier to use for OS (optional). - ...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "New Feature Additions" }
skypilot-org__skypilot-2136@6e9a8ab
skypilot-org/skypilot
Python
2,136
[Spot/UX]: Spot dashboard Added delay for ssh_command to initialize
<!-- Describe the changes in this PR --> Issue: #2123 User feedback > Small UX feedback: when I run sky spot dashboard, very often the page that opens first shows some error, and after a few seconds it automatically refreshes and loads. Slightly confusing user experience. Added a delay between ssh command and ...
2023-06-27T02:08:54Z
[spot dashboard/UI] Page failed to display in the first few seconds after opening User feedback > Small UX feedback: when I run sky spot dashboard, very often the page that opens first shows some error, and after a few seconds it automatically refreshes and loads. Slightly confusing user experience. I noticed tha...
This seems to be an issue of ssh_command needing some time to launch before the site is ready for webbrowser.open A naive solution might be to add a time.sleep() in between. @MaoZiming makes sense. Do you have time to put in a quick fix? Thanks.
[ { "body": "User feedback\r\n> Small UX feedback: when I run sky spot dashboard, very often the page that opens first shows some error, and after a few seconds it automatically refreshes and loads. Slightly confusing user experience.\r\n\r\n\r\nI noticed that too and not sure what's wrong.", "number": 2123, ...
7e76f4e0a72b0b24d8a478063713b050aa3b5b81
{ "head_commit": "6e9a8ab5a63e31d034edfb4d24b297abcba71245", "head_commit_message": "spot dashboard: Added delay for ssh_command to initialize", "patch_to_review": "diff --git a/sky/cli.py b/sky/cli.py\nindex c69e2c9fbe5..23556c80d62 100644\n--- a/sky/cli.py\n+++ b/sky/cli.py\n@@ -70,6 +70,7 @@\n from sky.utils i...
[ { "diff_hunk": "@@ -3736,6 +3737,7 @@ def spot_dashboard(port: Optional[int]):\n \n with subprocess.Popen(ssh_command, shell=True,\n start_new_session=True) as ssh_process:\n+ time.sleep(5) # Added delay for ssh_command to initialize.", "line": null, "original_line"...
4fd27435cb15e48767a243168cc6ac6553db067e
diff --git a/sky/cli.py b/sky/cli.py index c69e2c9fbe5..dc730cb3249 100644 --- a/sky/cli.py +++ b/sky/cli.py @@ -33,6 +33,7 @@ import subprocess import sys import textwrap +import time import typing from typing import Any, Dict, List, Optional, Tuple, Union import webbrowser @@ -3718,7 +3719,7 @@ def spot_dashboa...
{ "difficulty": "low", "estimated_review_effort": 2, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-2106@868fc0e
skypilot-org/skypilot
Python
2,106
[Spot] Event callback for adding notification
<!-- Describe the changes in this PR --> Issue #2000: A user would like to have a customizable event callback for the spot job, so that they can implement some notification when the spot job status changes. Example: ``` event_callback: send_slack_message $JOB_ID $JOB_STATUS ``` <!-- Describe the tests ran --> ...
2023-06-20T14:02:40Z
[Spot] Event callback for adding notification A user would like to have a customizable event callback for the spot jobs, so that they can implement some notification when the spot job status change. > Contributing one idea that would enable us to implement some custom notifications when spot jobs are killed / gone i...
For sending notifications (via Slack or Email), does SkyPilot have an existing mechanism? If not we need to register SkyPilot for Slack API and collect user info (slack id or email). > For sending notifications (via Slack or Email), does SkyPilot have an existing mechanism? If not we need to register SkyPilot for Slac...
[ { "body": "A user would like to have a customizable event callback for the spot jobs, so that they can implement some notification when the spot job status change.\r\n\r\n> Contributing one idea that would enable us to implement some custom notifications when spot jobs are killed / gone into error states: can w...
a9a9ab12867cbc625b81f2e611fdf42063e74fca
{ "head_commit": "868fc0e5c841ce9b1de2561b06bfabfc1abf2cfa", "head_commit_message": "running format.sh", "patch_to_review": "diff --git a/sky/spot/controller.py b/sky/spot/controller.py\nindex 1f4df44585e..7abccd7fe3a 100644\n--- a/sky/spot/controller.py\n+++ b/sky/spot/controller.py\n@@ -7,6 +7,7 @@\n import typ...
[ { "diff_hunk": "@@ -271,6 +276,35 @@ def _run_one_task(self, task_id: int, task: 'sky.Task') -> bool:\n task_id,\n recovered_time=recovered_time)\n \n+ def event_callback_func(self, task_id: int, state: str, comment: str = ''):\n+\n+ ...
a8b4135dae88c21fd55dcea8cb1522e152fee08c
diff --git a/sky/skylet/log_lib.py b/sky/skylet/log_lib.py index c4c4b0ca5ce..2c9c9bf4404 100644 --- a/sky/skylet/log_lib.py +++ b/sky/skylet/log_lib.py @@ -303,8 +303,8 @@ def add_ray_env_vars( def run_bash_command_with_log(bash_command: str, log_path: str, - ...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "New Feature Additions" }
skypilot-org__skypilot-2130@50f524b
skypilot-org/skypilot
Python
2,130
[Core] Launching cluster autodowned or manually terminated
<!-- Describe the changes in this PR --> Fixes #2129 We need to refactor the code in `execution` and the `_check_existing_cluster`, as the optimization and the provision are actually quite coupled, as raised by @WoosukKwon before. This PR is a fix that will unblock the issue #2129, but not the coupled issue. In ...
2023-06-24T23:03:05Z
[Python API] NoneType issue when re-launching a cluster terminated manually on the console An issue from our user, where our python API is not reliable for the cluster manually terminated or preempted on the cloud. To reproduce the issue: 1. `sky launch -c min --cloud gcp -i 120` 2. terminate the cluster on the co...
[ { "body": "An issue from our user, where our python API is not reliable for the cluster manually terminated or preempted on the cloud.\r\n\r\nTo reproduce the issue:\r\n1. `sky launch -c min --cloud gcp -i 120`\r\n2. terminate the cluster on the console manually\r\n3. `python -c 'import sky; sky.launch(sky.Task...
a5a3398f1c112223d987e651d668b8900e10c135
{ "head_commit": "50f524bbd0dc0df68ec1f15c880cd009ed562622", "head_commit_message": "remove uneccessary output", "patch_to_review": "diff --git a/sky/backends/cloud_vm_ray_backend.py b/sky/backends/cloud_vm_ray_backend.py\nindex 582b4cad6ce..f0030f8bead 100644\n--- a/sky/backends/cloud_vm_ray_backend.py\n+++ b/sk...
[ { "diff_hunk": "@@ -3701,6 +3704,20 @@ def _check_existing_cluster(\n if resources.cloud is not None else clouds.Cloud)\n task_cloud.check_cluster_name_is_valid(cluster_name)\n \n+ if to_provision is None:\n+ logger.info(\n+ f'The cluster {cluster_n...
d49366a264fbd310e7aebf44c373a6f41562f517
diff --git a/sky/backends/cloud_vm_ray_backend.py b/sky/backends/cloud_vm_ray_backend.py index b92c05511a4..b2f1bcae629 100644 --- a/sky/backends/cloud_vm_ray_backend.py +++ b/sky/backends/cloud_vm_ray_backend.py @@ -3680,7 +3680,8 @@ def run_on_head( @timeline.event def _check_existing_cluster( - ...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-1959@01ced15
skypilot-org/skypilot
Python
1,959
Add auto disk tier failover in Azure
<!-- Describe the changes in this PR --> This PR introduces auto disk tier failover in Azure and fixes one problem in #618 . <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the relevant ones): - [ ] Any manua...
2023-05-13T16:48:41Z
Failed to launch Azure gpunode GCP cpunode is unstable. I tried to `sky launch` 10 cpu nodes. But only 2 out of 10 succeeded. Errors happened during `ray up`. ``` [2/7] Processing file mounts Shared connection to 35.233.131.63 closed. ~/.sky/sky_wheels/ from /tmp/sky_wheels_weichiang/ Shared connection to...
OK... After I added `time.sleep(100)` https://github.com/sky-proj/sky/blob/f9b1d842f48d86610f4b3b2255a4f60a76a63e6d/sky/skylet/providers/gcp/node_provider.py#L193 and launched 10 cpu nodes. All of them becomes successful. Our above hypothesis of VM instability seems to be true. Not sure if it's related to https://...
[ { "body": "GCP cpunode is unstable. I tried to `sky launch` 10 cpu nodes. But only 2 out of 10 succeeded.\r\n\r\nErrors happened during `ray up`.\r\n```\r\n [2/7] Processing file mounts\r\nShared connection to 35.233.131.63 closed.\r\n ~/.sky/sky_wheels/ from /tmp/sky_wheels_weichiang/\r\nShared connection ...
fa7b478b935dbf89272ae0717808934716e503f2
{ "head_commit": "01ced156c949373688ba1040135be61d0ed4a109", "head_commit_message": "add failover to only specify instance type case", "patch_to_review": "diff --git a/sky/clouds/azure.py b/sky/clouds/azure.py\nindex 8227ab8e881..267cdda4f09 100644\n--- a/sky/clouds/azure.py\n+++ b/sky/clouds/azure.py\n@@ -245,6 ...
[ { "diff_hunk": "@@ -245,30 +245,46 @@ def make_deploy_resources_variables(\n }\n \n def get_feasible_launchable_resources(self, resources):\n+\n+ def failover_disk_tier(instance_type: str,\n+ disk_tier: Optional[str]) -> Optional[str]:\n+ ok, _ = Azure...
43a7e046eb3403992c364240bbaac82e2dc28c25
diff --git a/sky/clouds/azure.py b/sky/clouds/azure.py index 8227ab8e881..b5fb32ee924 100644 --- a/sky/clouds/azure.py +++ b/sky/clouds/azure.py @@ -245,6 +245,20 @@ def make_deploy_resources_variables( } def get_feasible_launchable_resources(self, resources): + + def failover_disk_tier(instance_...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-2098@ce6bf86
skypilot-org/skypilot
Python
2,098
[Core] Support launching a new cluster from an existing cluster's disk
<!-- Describe the changes in this PR --> Fixes #2050. This PR only supports AWS and GCP as an experimental feature and unblocks the users. ~~It is blocked by #2087~~ `sky launch --clone-disk-from existing-cluster -c new-cluster --cloud aws --region us-west-2 task.yaml` A problem with the image copying is th...
2023-06-16T22:25:14Z
Support migration of a stopped cluster to another region When a cluster is STOPPED and the resource of that cluster is not available in that region/zone, the user will not be able to restart the cluster and it would be hard to migrate to another region manually as the cluster may be a dev machine. To mitigate the pr...
[ { "body": "When a cluster is STOPPED and the resource of that cluster is not available in that region/zone, the user will not be able to restart the cluster and it would be hard to migrate to another region manually as the cluster may be a dev machine.\r\n\r\nTo mitigate the problem, we should provide a way to ...
d52070c39640ead9d9e1211f79f94d894fdc51d5
{ "head_commit": "ce6bf86dab33b76ea5afb0dd06c433f618271e6e", "head_commit_message": "fix", "patch_to_review": "diff --git a/docs/source/cloud-setup/cloud-permissions/aws.rst b/docs/source/cloud-setup/cloud-permissions/aws.rst\nindex 16f39ec3807..c30fa462cb5 100644\n--- a/docs/source/cloud-setup/cloud-permissions/...
[ { "diff_hunk": "@@ -1704,6 +1711,89 @@ def _query_cluster_status_via_cloud_api(\n return node_statuses\n \n \n+def check_can_clone_disk_and_override_task(\n+ cluster_name: str, target_cluster_name: Optional[str], task: 'task_lib.Task'\n+) -> Tuple['task_lib.Task', 'cloud_vm_ray_backend.CloudVmRayResource...
f6c77a5d2ca1d4206304cd9b36782913c8d27655
diff --git a/docs/source/cloud-setup/cloud-permissions/aws.rst b/docs/source/cloud-setup/cloud-permissions/aws.rst index 16f39ec3807..a212d5db16a 100644 --- a/docs/source/cloud-setup/cloud-permissions/aws.rst +++ b/docs/source/cloud-setup/cloud-permissions/aws.rst @@ -103,22 +103,36 @@ AWS accounts can be attached with...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "New Feature Additions" }
skypilot-org__skypilot-2066@886f3b8
skypilot-org/skypilot
Python
2,066
[Core] Fix log buffering issue
<!-- Describe the changes in this PR --> Fixes #2065. The original selectors implementation introduces buffer. We changed it to threading based implementation and it now fixes the buffering problem. Tried asyncio, but since the log_lib can be used in a thread and asyncio does not work well within a thread, we now...
2023-06-10T10:06:39Z
[Core] Logging buffered for run section When running the following yaml file, the `hi2` and `hi3` will only be printed after the `sleep` is finished. It can cause trouble for spot jobs debugging as the preemption can cause the latest log not showing up. ``` run: | echo hi echo hi2 echo hi3 sleep 12000 ...
This problem should be more critical than the progress bar
[ { "body": "When running the following yaml file, the `hi2` and `hi3` will only be printed after the `sleep` is finished. It can cause trouble for spot jobs debugging as the preemption can cause the latest log not showing up.\r\n```\r\nrun: |\r\n echo hi\r\n echo hi2\r\n echo hi3\r\n sleep 12000\r\n echo by...
47ee38c106dcdd47626f2fcd94e994c9389eab62
{ "head_commit": "886f3b830b427aecacb5d770da7d9cad8efb5551", "head_commit_message": "add comment", "patch_to_review": "diff --git a/sky/backends/cloud_vm_ray_backend.py b/sky/backends/cloud_vm_ray_backend.py\nindex ee98ac2fafb..c22ec0e8b9d 100644\n--- a/sky/backends/cloud_vm_ray_backend.py\n+++ b/sky/backends/clo...
[ { "diff_hunk": "@@ -26,84 +27,98 @@\n logger = sky_logging.init_logger(__name__)\n \n \n-def process_subprocess_stream(\n- proc,\n- log_path: str,\n- stream_logs: bool,\n- start_streaming_at: str = '',\n- end_streaming_at: Optional[str] = None,\n- skip_lines: Optional[List[str]] = None,\n- ...
26db596e77f46c500c5aee54e09155aaff82e351
diff --git a/sky/backends/cloud_vm_ray_backend.py b/sky/backends/cloud_vm_ray_backend.py index 5641835708b..d0356afcfe5 100644 --- a/sky/backends/cloud_vm_ray_backend.py +++ b/sky/backends/cloud_vm_ray_backend.py @@ -252,6 +252,8 @@ def add_prologue(self, job_id: int, is_local: bool = False) -> None: # FIX...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-1924@a9e83ef
skypilot-org/skypilot
Python
1,924
Added functionality to support specifying quantity of a resource in call to show-gpus
<!-- Describe the changes in this PR --> This PR fixes issue #1915. Previously, users were not able to specify the GPU quantity that they wanted. This PR allows users to do so by running sky show-gpus V100:4 to find the cloud providers and regions that offer 4 V100 GPUs. The output is now as follows: ``` sky show...
2023-05-02T00:29:26Z
[show-gpus] GPU count doesn't work with `sky show-gpus` We support resource specs like `V100:4`, but `sky show-gpus V100:4` doesn't work: ``` (base) ➜ ~ sky show-gpus V100:4 Resources 'V100:4' not found. Try 'sky show-gpus --all' to show available accelerators. ``` We should allow resource count when specifyi...
I'd like to take on this issue!
[ { "body": "We support resource specs like `V100:4`, but `sky show-gpus V100:4` doesn't work:\r\n\r\n```\r\n(base) ➜ ~ sky show-gpus V100:4\r\nResources 'V100:4' not found. Try 'sky show-gpus --all' to show available accelerators.\r\n```\r\n\r\nWe should allow resource count when specifying a GPU in `show-gpus`...
be313d00e6dc5f26cf2ddac0aa29b6e1a2d6c8a5
{ "head_commit": "a9e83ef643b98ddbe8f2a12590e801b4d4512078", "head_commit_message": "Clean code Trial3 - fixed repetitive code", "patch_to_review": "diff --git a/sky/cli.py b/sky/cli.py\nindex c5914171f24..46fde8c2b33 100644\n--- a/sky/cli.py\n+++ b/sky/cli.py\n@@ -2959,7 +2959,7 @@ def check():\n \n \n @cli.comm...
[ { "diff_hunk": "@@ -3063,16 +3063,35 @@ def _output():\n return\n \n # Show detailed accelerator information\n+\n+ name, quantity = accelerator_str.split(':')[0], None\n+ has_quantity = accelerator_str.count(':') > 0\n+ if has_quantity:\n+ quantity = int(a...
ba01c86100549273237eb538d54b737e4432f0a0
diff --git a/sky/cli.py b/sky/cli.py index c5914171f24..c8dd5376ed2 100644 --- a/sky/cli.py +++ b/sky/cli.py @@ -2959,7 +2959,7 @@ def check(): @cli.command() -@click.argument('gpu_name', required=False) +@click.argument('accelerator_str', required=False) @click.option('--all', '-a', ...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-1875@eb64396
skypilot-org/skypilot
Python
1,875
[Storage] Fix `sky storage delete` for externally deleted buckets
Closes #1322. If a bucket is deleted externally, then store.delete() now does a no-op (instead of creating and deleting like before). The logging has been updated. Now: ``` (base) ➜ repro git:(storage_deletionfix) ✗ SKYPILOT_DEBUG=0 sky storage delete -a Deleting all storage objects. # Shows spinner Deletin...
2023-04-17T03:08:03Z
[Storage] Confusing logging in `sky storage delete` When I am trying to delete a storage object with `sky storage delete skypilot-filemounts-*`, the logging is quite confusing showing the bucket is detected, then the bucket is created, and finally the bucket is deleted. ``` I 10-29 00:50:56 storage.py:402] Detected e...
I believe this is closed by #1323. New output: ``` (base) romilb@romilbx1yoga:~$ sky storage delete -a Deleting all storage objects. I 10-31 16:18:09 storage.py:1186] Deleted GCS bucket sky-romilb-gcsstore. I 10-31 16:18:25 storage.py:878] Deleted S3 bucket sky-romilb-s3store. ``` @romilbhardwaj On latest mast...
[ { "body": "When I am trying to delete a storage object with `sky storage delete skypilot-filemounts-*`, the logging is quite confusing showing the bucket is detected, then the bucket is created, and finally the bucket is deleted.\r\n```\r\nI 10-29 00:50:56 storage.py:402] Detected existing storage object, loadi...
508be85e4a2a0df1fa98f7fac63980ff6e24da3e
{ "head_commit": "eb64396213d0e71cd76a60a2dd5e76203f78e719", "head_commit_message": "lint", "patch_to_review": "diff --git a/sky/data/storage.py b/sky/data/storage.py\nindex a2355bac57a..bd801a7cafd 100644\n--- a/sky/data/storage.py\n+++ b/sky/data/storage.py\n@@ -53,6 +53,10 @@\n ' 2. If you are trying to c...
[ { "diff_hunk": "@@ -1455,9 +1507,11 @@ def _get_bucket(self) -> Tuple[StorageHandle, bool]:\n \"\"\"Obtains the GCS bucket.\n If the bucket exists, this method will connect to the bucket.\n \n- If the bucket does not exist, there are two cases:\n+ If the bucket does not exist, ther...
acca0b914c842a52aff271735fb078f719a0fdbd
diff --git a/sky/data/storage.py b/sky/data/storage.py index a2355bac57a..5225c283e9b 100644 --- a/sky/data/storage.py +++ b/sky/data/storage.py @@ -53,6 +53,10 @@ ' 2. If you are trying to connect to an existing bucket: make sure ' 'your cloud credentials have access to it.') +_BUCKET_EXTERNALLY_DELETED_D...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-1842@1767083
skypilot-org/skypilot
Python
1,842
GCP/provisioner: Handle the RESOURCE_NOT_FOUND error.
<!-- Describe the changes in this PR --> Fixes #1797. I've tried a few days and finally got a set_trace inside `need_ray_up()`, tested that the `re.search()` there works. This doesn't test things end-to-end, however. ``` sky launch -y --gpus A100-80GB:8 --cloud gcp --down -c dbg -r ``` <!-- Describe the te...
2023-04-08T17:23:39Z
RESOURCE_NOT_FOUND error in launching a GCP VM I met this error when launching an 8xA100 VM on GCP. The VM was initiated despite the error. ``` I 03-19 17:18:05 optimizer.py:617] == Optimizer == I 03-19 17:18:05 optimizer.py:628] Target: minimizing cost I 03-19 17:18:05 optimizer.py:640] Estimated cost: $29.4 / h...
Got a similar problem as well! Got reports that this happened again on spot A100:8 on GCP. @WoosukKwon @Michaelvll do you remember when you encountered the problem, were the on-demand VMs ssh-able? If so, we can retry `ray up` on this transient error. Got the same error again for launching on-demand A100-80GB:8, an...
[ { "body": "I met this error when launching an 8xA100 VM on GCP. The VM was initiated despite the error.\r\n\r\n```\r\nI 03-19 17:18:05 optimizer.py:617] == Optimizer ==\r\nI 03-19 17:18:05 optimizer.py:628] Target: minimizing cost\r\nI 03-19 17:18:05 optimizer.py:640] Estimated cost: $29.4 / hour\r\nI 03-19 17:...
ae5e9f6e5ddbfd65efb8a11e56ce99e0adac83c5
{ "head_commit": "1767083d9b8ed9dffd3b4be7096147e11469abb2", "head_commit_message": "GPC/provisioner: Mitigate RESOURCE_NOT_FOUND error for A100s.", "patch_to_review": "diff --git a/sky/backends/cloud_vm_ray_backend.py b/sky/backends/cloud_vm_ray_backend.py\nindex d5eb9ca59b0..be0577b8c0d 100644\n--- a/sky/backen...
[ { "diff_hunk": "@@ -657,6 +657,15 @@ def _update_blocklist_on_gcp_error(\n # {'code': 8, 'message': 'There is no more capacity in the zone \"europe-west4-a\"; you can try in another zone where Cloud TPU Nodes are offered (see https://cloud.google.com/tpu/docs/regions) [EID: 0x1bc8f9d790be914...
116083d228e86c0bb1f8a0faae8a2f529d489e5f
diff --git a/sky/backends/cloud_vm_ray_backend.py b/sky/backends/cloud_vm_ray_backend.py index d5eb9ca59b0..b983901d92d 100644 --- a/sky/backends/cloud_vm_ray_backend.py +++ b/sky/backends/cloud_vm_ray_backend.py @@ -657,6 +657,15 @@ def _update_blocklist_on_gcp_error( # {'code': 8, 'message': 'The...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-1865@45f1d89
skypilot-org/skypilot
Python
1,865
[R2] Fix bug on credential sync when AWS is disabled
<!-- Describe the changes in this PR --> This PR resolves #1844 When AWS is not enabled, `~/.aws/credentials` does not get sync to the remote cluster. As 1 of 2 credentials r2 requires is stored in `~/.aws/credentials` as well, r2 was not functioning when AWS was disabled. Hence, in this PR, it fixes to sync the `...
2023-04-15T08:27:59Z
R2 storage copy failed with return code 255. Tried the newly R2 storage support, following setup as described in the documentation got failed error ```failed with return code 255.``` tried to run command from the sky.exception.CommandError directly ``` mkdir -p ~/.sky/file_mounts/nli-datasets && aws --version >/d...
Thanks for the report @yudanta. This is a bug - we'll fix it. As a temporary workaround, if you have an AWS account, enabling [AWS support](https://skypilot.readthedocs.io/en/latest/getting-started/installation.html#aws) should let you use R2 on Lambda cloud. ## Reproduction I can confirm I can reproduce this bug w...
[ { "body": "Tried the newly R2 storage support, following setup as described in the documentation got failed error ```failed with return code 255.``` \r\n\r\ntried to run command from the sky.exception.CommandError directly\r\n```\r\nmkdir -p ~/.sky/file_mounts/nli-datasets && aws --version >/dev/null 2>&1 || pi...
f3acd9ec36c6cb0223c40e9ea5d1f5bc043c9520
{ "head_commit": "45f1d89863ace703e3d1e085ccec6f26612ac056", "head_commit_message": "fix", "patch_to_review": "diff --git a/sky/adaptors/cloudflare.py b/sky/adaptors/cloudflare.py\nindex 145665d701b..960a075f5a1 100644\n--- a/sky/adaptors/cloudflare.py\n+++ b/sky/adaptors/cloudflare.py\n@@ -5,11 +5,13 @@\n import...
[ { "diff_hunk": "@@ -119,3 +121,52 @@ def create_endpoint():\n endpoint = 'https://' + accountid + '.r2.cloudflarestorage.com'\n \n return endpoint\n+\n+\n+def check_credentials() -> Tuple[bool, Optional[str]]:\n+ \"\"\"Checks if the user has access credentials to this cloud.\"\"\"\n+\n+ hints = No...
ae7e05190d02e27cdbfd2061d0f9d8aa5bfdd92e
diff --git a/sky/adaptors/cloudflare.py b/sky/adaptors/cloudflare.py index 145665d701b..5f7e8024eeb 100644 --- a/sky/adaptors/cloudflare.py +++ b/sky/adaptors/cloudflare.py @@ -5,11 +5,13 @@ import functools import threading import os +from typing import Dict, Optional, Tuple boto3 = None botocore = None _sessi...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-1839@236ea8a
skypilot-org/skypilot
Python
1,839
Make rsync more resilient
<!-- Describe the changes in this PR --> Closes #1538. - `rsync` related errors are normally related to transient network issues, so this adds "to fix" information for the user in error messages - This also adds retrying to the `rsync` command to reduce the frequency of failures due to transient network issues ...
2023-04-06T04:14:35Z
UX: print "to fix" suggestions for rsync related errors Currently rsync related errors (e.g., transient network issues; we should look for repros) are not helpful because they don't display "how to fix" suggestions. In general, we should strive for including in each error message some "to fix" hints. Examples -...
A user met this error and didn't know it's a transient network connection issue. I'll draft a PR for this.
[ { "body": "Currently rsync related errors (e.g., transient network issues; we should look for repros) are not helpful because they don't display \"how to fix\" suggestions. \r\n\r\nIn general, we should strive for including in each error message some \"to fix\" hints.\r\n\r\nExamples -- we should audit all thes...
539046189bf808dfa1d8279dd99b79265cebe061
{ "head_commit": "236ea8a09c444b28111129b93b745b110c02cc53", "head_commit_message": "Fix formatting", "patch_to_review": "diff --git a/sky/backends/backend_utils.py b/sky/backends/backend_utils.py\nindex 5ef5b7b2903..500f4641baf 100644\n--- a/sky/backends/backend_utils.py\n+++ b/sky/backends/backend_utils.py\n@@ ...
[ { "diff_hunk": "@@ -292,6 +292,7 @@ def rsync(\n # Advanced options.\n log_path: str = os.devnull,\n stream_logs: bool = True,\n+ retry_cnt: int = 1,", "line": null, "original_line": 295, "original_start_line": null, "path": "sky/utils/command_runner.py", "star...
10ca8c75c77532391228b0d19df8986823c2f8bf
diff --git a/sky/backends/backend_utils.py b/sky/backends/backend_utils.py index 5ef5b7b2903..15c279a4c0c 100644 --- a/sky/backends/backend_utils.py +++ b/sky/backends/backend_utils.py @@ -1186,7 +1186,8 @@ def _sync_node(runner: 'command_runner.SSHCommandRunner') -> None: subprocess_utils.handle_returncod...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-1808@af2b0b5
skypilot-org/skypilot
Python
1,808
Allow fall back to account ID for AWS
<!-- Describe the changes in this PR --> Fixes #1807 We now allow the identity check fallback to AccountId if UserId does not match, because: 1.. In the case where multiple IAM users belong to a single root account, those users normally share the visibility of the VMs, so we do not need to identify them with eac...
2023-03-23T07:57:20Z
[AWS] Use Account ID instead of user id for identity check (we can make account id as the fallback option, which means we check the user id first and then check account id if the first one does not match and print warnings), reason In the case where multiple IAM users belong a single root account, those users normall...
[ { "body": " (we can make account id as the fallback option, which means we check the user id first and then check account id if the first one does not match and print warnings), reason\r\nIn the case where multiple IAM users belong a single root account, those users normally share the visibility of the VMs, so ...
67bd3cc402bdf6ebce02f4f9ab9ae56779c31a34
{ "head_commit": "af2b0b5498c2fb1d5690490b340b6fa6c8e56e1a", "head_commit_message": "Fix user identity", "patch_to_review": "diff --git a/sky/backends/backend_utils.py b/sky/backends/backend_utils.py\nindex 513615f7e45..45a7c00142a 100644\n--- a/sky/backends/backend_utils.py\n+++ b/sky/backends/backend_utils.py\n...
[ { "diff_hunk": "@@ -1744,7 +1744,11 @@ def check_owner_identity(cluster_name: str) -> None:\n if owner_identity is None:\n global_user_state.set_owner_identity_for_cluster(\n cluster_name, current_user_identity)\n- elif owner_identity != current_user_identity:\n+ else:\n+ as...
c91d206f0c1bd037f974eed0ba0208cd32ead89c
diff --git a/sky/backends/backend_utils.py b/sky/backends/backend_utils.py index 513615f7e45..2d023933e14 100644 --- a/sky/backends/backend_utils.py +++ b/sky/backends/backend_utils.py @@ -1744,7 +1744,27 @@ def check_owner_identity(cluster_name: str) -> None: if owner_identity is None: global_user_state....
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "New Feature Additions" }
skypilot-org__skypilot-2018@ddc23b5
skypilot-org/skypilot
Python
2,018
[Storage][Spot] Resolving failure to exclude local files for managed spot jobs using git --ignored outputs
This PR closes #1898 and opposed to implementing our own `.gitignore` filter in #1939 and using `rsync` from #1993 , this harness `git -C /path/to/source/dir status --ignored -s` to get a list of files/directories to be excluded from syncing `sky storage`. Added the testing structure and the files supposed to be ign...
2023-06-03T06:40:38Z
Spot launch doesn't respect `.gitignore` When launching an instance with `sky launch`, files included in the my `.gitignore` file are ignored during the rsync. However, when I launch a job with `sky spot launch`, it seems as though the files are included. Is this intended behavior?
Thanks for reporting the issue @keirp! This has been an issue for a while. The inconsistency comes from the cloud bucket’s rsync interface misalign with the system’s rsync. As we are using cloud bucket as the backend for the files uploading to the spot job, instead of using the system’s rsync for normal jobs, the probl...
[ { "body": "When launching an instance with `sky launch`, files included in the my `.gitignore` file are ignored during the rsync. However, when I launch a job with `sky spot launch`, it seems as though the files are included. Is this intended behavior?", "number": 1898, "title": "Spot launch doesn't res...
608d0c34ced75b881e5b35b36e416f56d2f685fc
{ "head_commit": "ddc23b53880087b929fa5f8dac78d113c5366f77", "head_commit_message": "gitignore filter using git", "patch_to_review": "diff --git a/sky/data/storage.py b/sky/data/storage.py\nindex 1c3420ad869..a5f25ad1512 100644\n--- a/sky/data/storage.py\n+++ b/sky/data/storage.py\n@@ -873,6 +873,75 @@ def add_if...
[ { "diff_hunk": "@@ -873,6 +873,75 @@ def add_if_not_none(key: str, value: Optional[Any]):\n return config\n \n \n+def format_gitignore_to_exclude_list(src_dir_path: str) -> List[str]:", "line": null, "original_line": 876, "original_start_line": null, "path": "sky/data/storage.py", "s...
5a6ca6c0b99b866312dd3875278de04311f386ee
diff --git a/sky/data/storage.py b/sky/data/storage.py index a9d6ce9f95c..66276e0f254 100644 --- a/sky/data/storage.py +++ b/sky/data/storage.py @@ -20,6 +20,7 @@ from sky.data import data_transfer from sky.data import data_utils from sky.data import mounting_utils +from sky.data import storage_utils from sky impor...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-1683@32b37ae
skypilot-org/skypilot
Python
1,683
Fix ssh config generation for multiple nodes
<!-- Describe the changes in this PR --> Fixes #1682 <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the relevant ones): - [x] Test snippets in #1682
2023-02-08T02:03:22Z
Generated ssh config in a wrong format for multiple nodes To reproduce: 1. `sky launch -c test --num-nodes 4 --cpus 2` 2. manually terminate 2 worker nodes 3. `sky launch -c test --num-nodes 4` The generated ssh config file (for the workers) becomes the following, which causes sytax error for `ssh` ``` # Added ...
[ { "body": "To reproduce:\r\n1. `sky launch -c test --num-nodes 4 --cpus 2`\r\n2. manually terminate 2 worker nodes\r\n3. `sky launch -c test --num-nodes 4`\r\n\r\nThe generated ssh config file (for the workers) becomes the following, which causes sytax error for `ssh`\r\n```\r\n# Added by sky (use `sky stop/dow...
dd72ad1e41794aa781a4a55d770734114a26d13a
{ "head_commit": "32b37ae831ef6ec94dd63b46a5c01b45995b469b", "head_commit_message": "fix ssh config gen", "patch_to_review": "diff --git a/sky/backends/backend_utils.py b/sky/backends/backend_utils.py\nindex 086cb67c4dc..dfe99718d3e 100644\n--- a/sky/backends/backend_utils.py\n+++ b/sky/backends/backend_utils.py\...
[ { "diff_hunk": "@@ -383,6 +383,7 @@ def _get_generated_config(cls, autogen_comment: str, host_name: str,\n Port 22\n {proxy}\n \"\"\".rstrip())\n+ codegen = codegen + '\\n' if not codegen.endswith('\\n') else codegen", "line": null, "original_line": 386, ...
1454353594f454402ea7db413ffd4413c569b3ab
diff --git a/sky/backends/backend_utils.py b/sky/backends/backend_utils.py index 086cb67c4dc..055e3daa57c 100644 --- a/sky/backends/backend_utils.py +++ b/sky/backends/backend_utils.py @@ -383,6 +383,7 @@ def _get_generated_config(cls, autogen_comment: str, host_name: str, Port 22 {proxy} ...
{ "difficulty": "low", "estimated_review_effort": 2, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-1713@fb29e84
skypilot-org/skypilot
Python
1,713
Fix show gpus when user does not have AWS credentials
<!-- Describe the changes in this PR --> Fixes #1710 <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the relevant ones): Without credential - [x] `sky show-gpus` (1.4 seconds; origin: 1.4 seconds) - [x] `s...
2023-02-22T07:13:04Z
`sky show-gpus` should not assume AWS credentials to fetch AZ mappings On a fresh docker install ``` pip install skypilot[lambda] sky show-gpus ``` shows ``` ... I 02-21 20:24:39 aws_catalog.py:53] Fetching availability zones mapping for AWS... --- Logging error --- .... File "/opt/conda/lib/python3.9/site...
[ { "body": "On a fresh docker install\r\n```\r\npip install skypilot[lambda]\r\nsky show-gpus\r\n```\r\nshows\r\n```\r\n...\r\nI 02-21 20:24:39 aws_catalog.py:53] Fetching availability zones mapping for AWS...\r\n--- Logging error ---\r\n....\r\n File \"/opt/conda/lib/python3.9/site-packages/sky/clouds/service_...
c94b43bb2a34538b5e88ad26081d1b498765ea0f
{ "head_commit": "fb29e84cdcd77fa180c4574ba4408f22e7510113", "head_commit_message": "fix test", "patch_to_review": "diff --git a/sky/clouds/service_catalog/aws_catalog.py b/sky/clouds/service_catalog/aws_catalog.py\nindex dfcba50848a..a6458a0a1e8 100644\n--- a/sky/clouds/service_catalog/aws_catalog.py\n+++ b/sky/...
[ { "diff_hunk": "@@ -44,6 +49,10 @@ def _apply_az_mapping(df: 'pd.DataFrame') -> 'pd.DataFrame':\n A dataframe with column 'AvailabilityZone' that's correctly replaced\n with the zone name (e.g. us-east-1a).\n \"\"\"\n+ global _az_mapping_applied\n+ if _az_mapping_applied:", "line":...
6f11747e87b0a38cddcc870c659fec56e21ed7b5
diff --git a/sky/cli.py b/sky/cli.py index bdb246f15a9..c931d059b8b 100644 --- a/sky/cli.py +++ b/sky/cli.py @@ -2724,6 +2724,7 @@ def check(): ('The region to use. If not specified, shows accelerators from all regions.' ), ) +@service_catalog.use_default_catalog @usage_lib.entrypoint def show_gpus( ...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-1667@88b151f
skypilot-org/skypilot
Python
1,667
[Spot] Disallow `sky down` spot controller when in-progress spot jobs exist
<!-- Describe the changes in this PR --> Fixes #1666 With this PR the following message will show if `sky down` the spot controller with in-progress spot jobs: <img width="799" alt="image" src="https://user-images.githubusercontent.com/6753189/216758225-2654732a-ea1d-4800-8b84-1e5fa2927bcc.png"> Tested: - ...
2023-02-04T08:49:43Z
[Spot] Disallow terminating the spot cluster when there is spot jobs running Requested by user: We should ban the termination of the spot controller (not even showing the prompt) when there is spot job running to avoid spot resources leakage. We should also make the termination of the spot controller harder, by a...
[ { "body": "Requested by user:\r\n\r\nWe should ban the termination of the spot controller (not even showing the prompt) when there is spot job running to avoid spot resources leakage.\r\n\r\nWe should also make the termination of the spot controller harder, by asking the user to type `I understand` etc instead ...
2b4b465270dcd19d487ec111efdf72a8b2b90166
{ "head_commit": "88b151f6a7dc952884fe1f733b06ee02b450c5ac", "head_commit_message": "address comments", "patch_to_review": "diff --git a/sky/cli.py b/sky/cli.py\nindex 10288245c55..8d5c97fabfe 100644\n--- a/sky/cli.py\n+++ b/sky/cli.py\n@@ -2200,7 +2200,7 @@ def down(\n purge=purge)\n \...
[ { "diff_hunk": "@@ -2316,7 +2331,7 @@ def _down_or_stop_clusters(\n # TODO(zhwu): We can only have one reserved cluster (spot\n # controller).\n assert len(reserved_clusters) == 1, reserved_clusters\n- _hint_for_down_spot_controller(reserved_cluster...
5353f62d8ed7afa5104a7b38751f56ad3b214ae2
diff --git a/sky/cli.py b/sky/cli.py index 10288245c55..bdb246f15a9 100644 --- a/sky/cli.py +++ b/sky/cli.py @@ -2200,7 +2200,7 @@ def down( purge=purge) -def _hint_for_down_spot_controller(controller_name: str): +def _hint_or_raise_for_down_spot_controller(controller_name: str): # ...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-1632@0bce7d4
skypilot-org/skypilot
Python
1,632
[Setup] Remove the dependency for typing_extensions
<!-- Describe the changes in this PR --> Fixes #1631 <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the relevant ones): - [x] `pip install .`
2023-01-25T18:12:13Z
ModuleNotFoundError: No module named 'typing_extensions' Good morning sky team, An `ModuleNotFoundError` after update skypilot yesterday. ```{shell} $ sky status -r Traceback (most recent call last): File "/Users/hanqingliu/mambaforge/bin/sky", line 5, in <module> from sky.cli import cli File "/Users...
Thank you for finding this! Could you try to run `pip install typing-extensions` to install that dependency?
[ { "body": "Good morning sky team,\r\n\r\nAn `ModuleNotFoundError` after update skypilot yesterday.\r\n\r\n```{shell}\r\n$ sky status -r\r\nTraceback (most recent call last):\r\n File \"/Users/hanqingliu/mambaforge/bin/sky\", line 5, in <module>\r\n from sky.cli import cli\r\n File \"/Users/hanqingliu/mamba...
4dda5ee3fe404e86b76a8ba8b5d03e2166033650
{ "head_commit": "0bce7d446ec561fe061d6e6a1256a4a83b56f0ab", "head_commit_message": "Add type-extensions to setup.py", "patch_to_review": "diff --git a/sky/setup_files/setup.py b/sky/setup_files/setup.py\nindex 4857f0e82fe..26b43d47950 100644\n--- a/sky/setup_files/setup.py\n+++ b/sky/setup_files/setup.py\n@@ -83...
[ { "diff_hunk": "@@ -83,6 +83,7 @@ def parse_readme(readme: str) -> str:\n 'ray[default]>=1.9.0,<=2.2.0',\n 'rich',\n 'tabulate',\n+ 'typing-extensions',", "line": null, "original_line": 86, "original_start_line": null, "path": "sky/setup_files/setup.py", "start_line": null, ...
6a2ebe3b0ee137f59f89cff3bbf3ec953d5dc28d
diff --git a/sky/data/storage.py b/sky/data/storage.py index 69dd1e45d8d..513a173cc11 100644 --- a/sky/data/storage.py +++ b/sky/data/storage.py @@ -5,7 +5,6 @@ import time import typing from typing import Any, Dict, Optional, Tuple, Type, Union, List -from typing_extensions import TypeAlias import urllib.parse ...
{ "difficulty": "low", "estimated_review_effort": 1, "problem_domain": "Dependency Updates & Env Compatibility" }
skypilot-org__skypilot-1910@3f212d2
skypilot-org/skypilot
Python
1,910
Add docker support for SkyPilot
<!-- Describe the changes in this PR --> This PR enable user-provided docker image on VM. TODO: - Test SSHProxyCommand on real-world usecase <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the relevant ones)...
2023-04-28T16:26:36Z
[UX] Container support for task We have received multiple requests from our users for the native container support in the task yaml, so that they don't need to manually run the `docker pull` and `docker run` in the `setup` and `run` sections separately.
A blocker for one of our potential user. Does this mean they would like the `run` commands to be executed within the docker container as well? Yes, I suppose both setup and run sections will be run in the container, if that is posible.
[ { "body": "We have received multiple requests from our users for the native container support in the task yaml, so that they don't need to manually run the `docker pull` and `docker run` in the `setup` and `run` sections separately.", "number": 1681, "title": "[UX] Container support for task" } ]
8dbdc89fb0593e3f0fd6f27eda8a39ffea9bd336
{ "head_commit": "3f212d2c6599a7698759990467d9b3ec2e39640f", "head_commit_message": "remove error merging", "patch_to_review": "diff --git a/docs/source/reference/yaml-spec.rst b/docs/source/reference/yaml-spec.rst\nindex 66d264189fe..3ed0c671831 100644\n--- a/docs/source/reference/yaml-spec.rst\n+++ b/docs/sourc...
[ { "diff_hunk": "@@ -871,6 +878,10 @@ def write_cluster_config(\n f'open(os.path.expanduser(\"{constants.SKY_REMOTE_RAY_PORT_FILE}\"), \"w\"))\\''\n )\n \n+ # pylint: disable=import-outside-toplevel\n+ from sky.backends.docker_utils import \\\n+ DEFAULT_DOCKER_CONTAINER_NAME", "line"...
56df4f6e785276bb1f5f6c9c28494bebbb62cf34
diff --git a/docs/source/reference/yaml-spec.rst b/docs/source/reference/yaml-spec.rst index fdc164af586..f7ff72a0c6c 100644 --- a/docs/source/reference/yaml-spec.rst +++ b/docs/source/reference/yaml-spec.rst @@ -109,8 +109,15 @@ Available fields: tpu_vm: False # False to use TPU nodes (the default); True to ...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "New Feature Additions" }
skypilot-org__skypilot-1622@2c08184
skypilot-org/skypilot
Python
1,622
[Resources] Add `cpus` in resource specification
<!-- Describe the changes in this PR --> Closes #387 This PR introduces the ability for users to specify resource requirements in terms of the number of vCPUs in the `Resources` class and the `resources` section in the SkyPilot yaml. **Semantics** 1. If the instance type and accelerators are not specified, SkyP...
2023-01-24T08:49:52Z
Resource Specification for # of CPUs Sky currently supports users requesting X amount of GPUs/TPUs, e.g. ``` resources: accelerators: V100: 8 ``` Users should also be able to specify how many CPU cores they want, e.g. ``` resources: cores: 16 accelerators: V100: 8 ``` Additional changes are ...
We knew this needs to be happen at some point. Question is: why does Daniel want it? What would this achieve for him? Does he want it or does he need it right now? Does he want it for (i) the initial cluster provisioning; or (ii) constraining a task's cpu use? Based on what I saw, I think both. (i) Initial cluste...
[ { "body": "Sky currently supports users requesting X amount of GPUs/TPUs, e.g. \r\n```\r\nresources:\r\n accelerators:\r\n V100: 8\r\n```\r\nUsers should also be able to specify how many CPU cores they want, e.g.\r\n```\r\nresources:\r\n cores: 16\r\n accelerators:\r\n V100: 8\r\n```\r\nAdditional chan...
a9c194a4ab3d75eea9e388dddc92ff07dfe6fcf0
{ "head_commit": "2c081845e45d19f4ec857bc0024511d3cae527ae", "head_commit_message": "Fix", "patch_to_review": "diff --git a/sky/cli.py b/sky/cli.py\nindex 8d504a0ce59..7a757df0c0f 100644\n--- a/sky/cli.py\n+++ b/sky/cli.py\n@@ -174,6 +174,13 @@ def _interactive_node_cli_command(cli_func):\n ...
[ { "diff_hunk": "@@ -148,6 +163,41 @@ def test_clouds_not_enabled(monkeypatch):\n enabled_clouds=[sky.AWS()])\n \n \n+def test_instance_type_mismatches_cpus(monkeypatch):\n+ bad_instance_and_cpus = [\n+ # Actual: 8\n+ ('m6i.2xlarge', 4),\n+ # Actual: 2\n+ ...
4ea39d7ffdd3038bc5c575a43af5850caf86197f
diff --git a/docs/source/reference/yaml-spec.rst b/docs/source/reference/yaml-spec.rst index 586e0366cbc..1465deff853 100644 --- a/docs/source/reference/yaml-spec.rst +++ b/docs/source/reference/yaml-spec.rst @@ -47,6 +47,12 @@ Available fields: # Format: <name>:<count> (or simply <name>, short for a count of 1)...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "New Feature Additions" }
skypilot-org__skypilot-1637@d4a828b
skypilot-org/skypilot
Python
1,637
[Autostop/UX] Better UX when the cluster is partially stopped due to autostopping
<!-- Describe the changes in this PR --> Fixes #1633 Fixes the confusion of cluster being in `INIT` status, when it is being autostopped and is partially stopped. Changes: 1. keep the autostop column in the cluster table, when the cluster is partially stopped 2. Print a hint that the cluster is being autostop...
2023-01-26T19:41:42Z
[Autostop] `sky status --refresh` after autostop fails for multi-node on GCP ## Reproduction ``` # launch multinode gcp cluster sky launch -y -d -c test --num-nodes 2 --cloud gcp tests/test_yamls/minimal.yaml # Try sky status --refresh. This works fine. sky status test --refresh # Schedule autostop sky autos...
Just to add, in `test_autostop` the state is shown to be at `INIT` instead of `UP` or `STOPPED`, which causes the test to fail. [Logs here](https://gist.github.com/romilbhardwaj/4be928e440d914d66bccd1045d7a9e14). Partially addressed by #1634 and #1638, removing P0 label
[ { "body": "## Reproduction\r\n```\r\n# launch multinode gcp cluster\r\nsky launch -y -d -c test --num-nodes 2 --cloud gcp tests/test_yamls/minimal.yaml\r\n\r\n# Try sky status --refresh. This works fine.\r\nsky status test --refresh\r\n\r\n# Schedule autostop\r\nsky autostop -y test -i 1\r\n\r\n# Wait for autos...
e535f29546fe905eb8d651a08fc436fbfce68ff1
{ "head_commit": "d4a828bb0e47f8d3daf54913be9074b1cfad49ce", "head_commit_message": "address comments", "patch_to_review": "diff --git a/sky/backends/backend_utils.py b/sky/backends/backend_utils.py\nindex b83dc5d8a99..6fc1b8b1a6b 100644\n--- a/sky/backends/backend_utils.py\n+++ b/sky/backends/backend_utils.py\n@...
[ { "diff_hunk": "@@ -128,7 +128,7 @@ def _run(self):\n def _stop_cluster(self, autostop_config):\n if (autostop_config.backend ==\n cloud_vm_ray_backend.CloudVmRayBackend.NAME):\n- autostop_lib.set_is_autostopping()\n+ autostop_lib.set_autostopping_indicator()", ...
bfeca46835d6efa3112b1f7164bcf9f538fc8787
diff --git a/sky/backends/backend_utils.py b/sky/backends/backend_utils.py index b83dc5d8a99..65124347889 100644 --- a/sky/backends/backend_utils.py +++ b/sky/backends/backend_utils.py @@ -1950,24 +1950,34 @@ def _update_cluster_status_no_lock( # abnormal. # # An abnormal cluster will transition to I...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-1718@1b51ae2
skypilot-org/skypilot
Python
1,718
[Lambda Cloud] Multinode support
This pr extends Lambda Cloud support to multi-node clusters. Some notes: - The tag file scheme has been changed (because the head node calls `ray start` to launch the children nodes). - Generic smoke tests that required multi-node have the `@no_lambda_labs` descriptor removed. - We use `ssh` to get `internal_ip`...
2023-02-23T22:35:48Z
[Lambda Cloud] Handle `Ctrl-C` during launch Running `sky launch --cloud lambda` followed immediately by `Ctrl-C` and `sky down -a` will yield: ``` E 02-06 16:45:41 cloud_vm_ray_backend.py:3042] **** STDERR **** E 02-06 16:45:41 cloud_vm_ray_backend.py:3042] 2023-02-06 16:45:39,483 INFO util.py:335 -- setting max w...
[ { "body": "Running `sky launch --cloud lambda` followed immediately by `Ctrl-C` and `sky down -a` will yield:\r\n```\r\nE 02-06 16:45:41 cloud_vm_ray_backend.py:3042] **** STDERR ****\r\nE 02-06 16:45:41 cloud_vm_ray_backend.py:3042] 2023-02-06 16:45:39,483 INFO util.py:335 -- setting max workers for head node...
464b5db86ee1d4bd7b96862b6e5a44e2f76140b8
{ "head_commit": "1b51ae2b04f6defee01d9dc0ffdfe0d680f1697c", "head_commit_message": "Increase timeout again and nit", "patch_to_review": "diff --git a/sky/backends/backend_utils.py b/sky/backends/backend_utils.py\nindex 2a14c40b287..40781294097 100644\n--- a/sky/backends/backend_utils.py\n+++ b/sky/backends/backe...
[ { "diff_hunk": "@@ -16,12 +16,14 @@\n NODE_KIND_WORKER,\n NODE_KIND_HEAD,\n )\n-from ray.autoscaler._private.util import hash_launch_conf\n from sky.skylet.providers.lambda_cloud import lambda_utils\n-from sky.utils import common_utils\n+from sky import authentication as auth\n+from sky.utils import com...
edf030a72019f2dac44bb56d37ccac5a0d8cd957
diff --git a/sky/backends/backend_utils.py b/sky/backends/backend_utils.py index 869032c1c0c..3e36e050212 100644 --- a/sky/backends/backend_utils.py +++ b/sky/backends/backend_utils.py @@ -1692,11 +1692,13 @@ def _query_status_lambda( 'terminated': None, } # TODO(ewzeng): filter by hash_filter_string...
{ "difficulty": "medium", "estimated_review_effort": 4, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-1617@69ec1d2
skypilot-org/skypilot
Python
1,617
[UX] Make `sky logs <nonexistent job id>` not error out.
<!-- Describe the changes in this PR --> Fixes #1607. <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the relevant ones): - [x] Any manual or new tests for this PR (please specify below) ``` » sky logs sky-...
2023-01-21T17:51:21Z
[UX] sky logs throws unnecessary errors when job id doesn't exist. `sky logs` throws errors when the job id doesn't exist. We should catch the error instead and show `Job not found` or things like that. ``` (sky-tmp) weichiang@mbp sky % sky logs cpu 1 Tailing logs of job 1 on cluster 'cpu'... Warning: Permanently a...
[ { "body": "`sky logs` throws errors when the job id doesn't exist. We should catch the error instead and show `Job not found` or things like that.\r\n```\r\n(sky-tmp) weichiang@mbp sky % sky logs cpu 1\r\nTailing logs of job 1 on cluster 'cpu'...\r\nWarning: Permanently added '35.202.33.57' (ED25519) to the lis...
9fca91b31c0683f62f44d8f3b6ec22d42b8e648f
{ "head_commit": "69ec1d24fd1fbb148eceb8e21c115e97b1b9e1da", "head_commit_message": "Make `sky logs <nonexistent job id>` not error out.", "patch_to_review": "diff --git a/sky/skylet/job_lib.py b/sky/skylet/job_lib.py\nindex 62909440621..17aa608fb54 100644\n--- a/sky/skylet/job_lib.py\n+++ b/sky/skylet/job_lib.py...
[ { "diff_hunk": "@@ -670,9 +668,10 @@ def tail_logs(cls,\n 'else job_lib.get_latest_job_id()',\n 'run_timestamp = job_lib.get_run_timestamp(job_id)',\n (f'log_dir = os.path.join({constants.SKY_LOGS_DIRECTORY!r}, '\n- 'run_timestamp)'),\n+ '\"\" if run_t...
c14999303ba36c668b6387e0b81c62d4ca9b97e1
diff --git a/sky/skylet/job_lib.py b/sky/skylet/job_lib.py index 62909440621..2946bcc8aa9 100644 --- a/sky/skylet/job_lib.py +++ b/sky/skylet/job_lib.py @@ -362,17 +362,14 @@ def _get_jobs_by_ids(job_ids: List[int]) -> List[Dict[str, Any]]: def update_job_status(job_owner: str, job_ids: List[int...
{ "difficulty": "low", "estimated_review_effort": 2, "problem_domain": "Bug Fixes" }
skypilot-org__skypilot-1588@c32ff5f
skypilot-org/skypilot
Python
1,588
UX: hint for unmanaged spot; spot queue logging.
<!-- Describe the changes in this PR --> Fixes #1553. - Hint for unmanaged spot - Some logging changes in `sky spot queue [-s]` <!-- Describe the tests ran --> <!-- Unit tests (tests/test_*.py) are part of GitHub CI; below are tests that launch on the cloud. --> Tested (run the relevant ones): `spot ...
2023-01-13T17:29:49Z
UX: if unmanaged spot is used, print a hint that only managed spot automatically recovers If we detected unmanaged spot is used, we should print a hint a la - Unmanaged spot cluster is being created, which does not automatically recover from preemptions. - To have auto recovery, use managed spot jobs instead: CLI `s...
[ { "body": "If we detected unmanaged spot is used, we should print a hint a la\r\n- Unmanaged spot cluster is being created, which does not automatically recover from preemptions. \r\n- To have auto recovery, use managed spot jobs instead: CLI `sky spot launch`, Python `sky.spot_launch()`.", "number": 1553, ...
ad6296b71f8c6dde35ed47083e920606fb4e1b00
{ "head_commit": "c32ff5fc58eac2f624329a1b22f33c04e7b9d2ff", "head_commit_message": "Pylint", "patch_to_review": "diff --git a/sky/cli.py b/sky/cli.py\nindex e4c9eff8f46..5cb7478004e 100644\n--- a/sky/cli.py\n+++ b/sky/cli.py\n@@ -2995,26 +2995,43 @@ def spot_queue(all: bool, refresh: bool, skip_finished: bool):\...
[ { "diff_hunk": "@@ -167,6 +170,16 @@ def _execute(\n raise ValueError(\n 'Spot recovery is specified in the task. To launch the '\n 'managed spot job, please use: sky spot launch')\n+ if task.use_spot and not _is_launched_by_spot_controller:\n+ yellow = colo...
5802219e59dcdf6789fb41905711efd4c05540a3
diff --git a/sky/cli.py b/sky/cli.py index e4c9eff8f46..b80724aecc2 100644 --- a/sky/cli.py +++ b/sky/cli.py @@ -2995,26 +2995,48 @@ def spot_queue(all: bool, refresh: bool, skip_finished: bool): watch -n60 sky spot queue """ click.secho('Fetching managed spot job statuses...', fg='yellow') + no_job...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "New Feature Additions" }
skypilot-org__skypilot-1572@1712726
skypilot-org/skypilot
Python
1,572
Various UX Fixes
Addresses #1565, #1570, #1482 and various other bug fixes/small additions: * Add move hourly price computation from `status_utils` to `get_hourly_price` method so it can be accessed after `core.status` * Added confirmation before cancelling jobs * Remove confirmation when `--yes` is specified for `sky down` of spo...
2023-01-05T10:27:25Z
[Spot/UX] Need a way to only show the RUNNING spot jobs Request from our user: > Is there a way to clear jobs that are finished (either "SUCCEEDED" or "FAILED")? [Job] Inconsistent job cancel behavior Our user reported that the `sky cancel` does not require confirmation but the `sky spot cancel` requires. The user...
From the discussion, seems like one option is to only show the `RUNNING` jobs by default. We may also want to add some hint message like ``` Note: Only RUNNING jobs are displayed, run sky spot queue --all to show the entire history ``` Two other possibilities: - `sky queue` already provides `-s, --skip-finished Sh...
[ { "body": "Request from our user:\r\n> Is there a way to clear jobs that are finished (either \"SUCCEEDED\" or \"FAILED\")?\r\n", "number": 1482, "title": "[Spot/UX] Need a way to only show the RUNNING spot jobs" }, { "body": "Our user reported that the `sky cancel` does not require confirmation...
f84fb0d5b93ee2f9453b47fe97b03edfa549360e
{ "head_commit": "1712726ba1ce63268d669f5c2e1de46b0c89de00", "head_commit_message": "formatting", "patch_to_review": "diff --git a/sky/backends/cloud_vm_ray_backend.py b/sky/backends/cloud_vm_ray_backend.py\nindex fd22646705f..2281b147938 100644\n--- a/sky/backends/cloud_vm_ray_backend.py\n+++ b/sky/backends/clou...
[ { "diff_hunk": "@@ -651,7 +652,7 @@ def spot_queue(refresh: bool) -> List[Dict[str, Any]]:\n 'end_at': (float) timestamp of end,\n 'duration': (float) duration in seconds,\n 'retry_count': int Number of retries,\n- 'status': sky.JobStatus status of ...
94248fbacdf3fceb8545adc71b3e039deb91e812
diff --git a/sky/backends/cloud_vm_ray_backend.py b/sky/backends/cloud_vm_ray_backend.py index fd22646705f..2281b147938 100644 --- a/sky/backends/cloud_vm_ray_backend.py +++ b/sky/backends/cloud_vm_ray_backend.py @@ -1870,6 +1870,11 @@ def external_ips(self, return [ips[1] for ips in self.stable_intern...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "New Feature Additions" }
skypilot-org__skypilot-1532@8d39ecd
skypilot-org/skypilot
Python
1,532
Fix sky start for AMI without conda
<!-- Describe the changes in this PR --> Describe the changes in this PR: Fixes https://github.com/skypilot-org/skypilot/issues/1517 and add one smoke test to cover this case in which an AMI without conda is used. The bug can be reproduced by the following yaml. ``` resources: cloud: aws instance_type: g...
2022-12-17T08:17:08Z
Error when restarting cluster Hi, I'm running into the following error when restarting a cluster: `ModuleNotFoundError: No module named 'sky'`. Steps to reproduce: 1. `sky launch -C my_cluster config.yaml` 2. `sky stop my_cluster` 3. `sky start my_cluster` The instance is successfully provisioned on AWS, and I a...
Thank you for reporting the issue @pschafhalter! The SkyPilot runtime seems to have been wrongly set up on your remote cluster. Do you mind sharing the resources field of your `config.yaml` or the `image_id` you are using for the cluster to make it easier for us to debug the problem? PS, I tried the latest master b...
[ { "body": "Hi, I'm running into the following error when restarting a cluster: `ModuleNotFoundError: No module named 'sky'`.\r\nSteps to reproduce:\r\n1. `sky launch -C my_cluster config.yaml`\r\n2. `sky stop my_cluster`\r\n3. `sky start my_cluster`\r\n\r\nThe instance is successfully provisioned on AWS, and I ...
e9b378e63296e0a545bc83e6f54e05e4a0afd765
{ "head_commit": "8d39ecd613246517fdbdff5265c767e637855478", "head_commit_message": "rename", "patch_to_review": "diff --git a/examples/tests/test_yamls/no_conda_ami.yaml b/examples/tests/test_yamls/no_conda_ami.yaml\nnew file mode 100644\nindex 00000000000..5aae1f70a55\n--- /dev/null\n+++ b/examples/tests/test_y...
[ { "diff_hunk": "@@ -0,0 +1,9 @@\n+resources:\n+ cloud: aws\n+ instance_type: g4dn.xlarge\n+ region: us-west-2\n+ image_id: ami-0fe5af21074ad2a10 # Deep learning AMI with CUDA 11.6", "line": null, "original_line": 5, "original_start_line": null, "path": "examples/tests/test_yamls/no_conda_am...
c3bf4c3a611f9f06227b938320f04cbc3a28278b
diff --git a/examples/tests/test_yamls/no_conda_ami.yaml b/examples/tests/test_yamls/no_conda_ami.yaml new file mode 100644 index 00000000000..c969d44118d --- /dev/null +++ b/examples/tests/test_yamls/no_conda_ami.yaml @@ -0,0 +1,9 @@ +resources: + cloud: aws + instance_type: g4dn.xlarge + region: us-west-2 + image...
{ "difficulty": "medium", "estimated_review_effort": 3, "problem_domain": "Bug Fixes" }